2018 Vol. 47, No. 2
column
- Invited column-deep learning and its application
- Infrared technology and application
- Laser technology and application
- Laser radar technology
- Terahertz technology and application
- Optical communication and optical sensing
- Photoelectric device and microsystem
- Photoelectric devices and microsystem
- Photoelectric measurement
- Information acquisition and identification
2018, 47(2): 203001.
doi: 10.3788/IRLA201847.0203001
In the task of close range pedestrian detection, the balance of the precision and speed were of great significance to the practical application of the detection algorithm. In order to detect the close range target quickly and accurately, a pedestrian detection algorithm based on fused fully convolutional network was proposed. Firstly, a fully convolutional detection network was used to detect the target in the image, and a series of candidate bounding boxes were obtained. Secondly, pixel level classification results of the image were obtained by using a semantic segmentation network with weakly supervised training. Finally, the candidate bounding boxes and the pixel level classification results were fused to complete the detection. The experimental results show that the algorithm has good performance in both the speed and the precision of detection.
2018, 47(2): 203002.
doi: 10.3788/IRLA201847.0203002
Convolution Neural Networks (CNN) and Recurrent Neural Networks (RNN) had developed rapidly in the fields of image classification, computer vision, natural language process, speech recognition, machine translation and semantic analysis, which caused researchers' close attention to computers' automatic generation of image interpretation. At present, the main problems in image description were sparse input text data, over-fitting of the model, difficult convergence of the model loss function, and so on. In this paper, NIC was used as a baseline model. For data sparseness, one-hot text in the baseline model was changed and word2vec was used to map the text. To prevent over-fitting, regular items were added to the model and Dropout technology was used. In order to make innovations in word order memory, the associative memory unit GRU for text generation was used. In experiment, the AdamOptimizer optimizer was used to update parameters iteratively. The experimental results show that the improved model parameters are reduced and the convergence speed is significantly faster, the loss function curves are smoother, the maximum loss is reduced to 2.91, and the model accuracy rate increases by nearly 15% compared with the NIC. Experiments validate that the use of word2vec to map text in the model obviously alleviates the data sparseness problem. Adding regular items and using Dropout technology could effectively prevent over-fitting of the model. The introduction of associative memory unit GRU could greatly reduce the model trained parameters and speed up the algorithm of convergence rate, improve the accuracy of the entire model.
2018, 47(2): 203003.
doi: 10.3788/IRLA201847.0203003
Generative adversarial networks had shown promising potential in conditional image generation. It seemed that the GANs were particularly suitable for use in image super-resolution reconstruction. However, there was a shortcoming of excessive smoothness and lack of high frequency detail information for the reconstructed SR images by using GANs. Aiming at resolving the problem that the method of single image super-resolution reconstruction ignored the spatio-temporal relationship between image frames, a method of multiframe infrared image super-resolution reconstruction based on generative adversarial networks (M-GANs) was proposed in this paper. Firstly, motion compensation was proposed for registration low resolution image frames; Secondly, a weight representation convolutional layer was performed to calculate the weight transfer; Finally, the generative adversarial network was used to reconstruct the high resolution image. Experimental results demonstrate that the proposed method surpass current state-of-the-art performance of both subjective and objective evaluation.
2018, 47(2): 203004.
doi: 10.3788/IRLA201847.0203004
Perceptual aliasing and perceptual variability caused by drastically appearance changing in the scene bring great challenge to visual place recognition. Many existing visual place recognition methods using CNN directly adopted the distance of the CNN features and set thresholds to measure the similarity between the two images, which had shown a poor performance when drastically appearance changing in the scene. A novel multi-level feature difference map based visual place recognition method was proposed. Firstly, a CNN pretrained on scene-centric dataset was adopted to extract features for perceptually different images of same place and aliased images of different places. Then, according to the different properties of different CNN layers, multi-level feature difference map was constructed on the multi-level CNN features to represent the difference between the two images. Finally, visual place recognition was regarded as a binary classification task. The feature difference maps were used to train a new CNN classification model for determining whether the two images are from the same place. Experimental results demonstrated that the feature difference map constructed by multi-level CNN features can well represent the difference between two images, and the proposed method can effectively overcome perceptual aliasing and perceptual variability, and achieve a better recognition performance when drastically appearance changing in the scene.
2018, 47(2): 203005.
doi: 10.3788/IRLA201847.0203005
Texture synthesis is a hot research topic in the fields of computer graphics, vision, and image processing. Traditional texture synthesis methods are generally achieved by extracting effective feature patterns or statistics and generating random images under the constraint of the feature information. Generative adversarial networks (GANs) is a new type of deep network. It can randomly generate new data of the same distribution as the observed data by training generator and discriminator in an adversarial learning mechanism. Inspired by this point, a texture synthesis method based on GANs was proposed. The advantage of the algorithm was that it could generate more realistic texture images without iteration; the generated images were visually consistent with the observed texture image and also had randomness. A series of experiments for random texture and structured texture synthesis verify the effectiveness of the proposed algorithm.
2018, 47(2): 203006.
doi: 10.3788/IRLA201847.0203006
In view of the complicated background of the fire image, the complicated process of extracting the artificial feature, the poor generalization ability of the fire image, the low accuracy, false alarm rate, missing rate, the novel method for detecting fire images of multilinear principal component analysis (MPCA) was presented in the paper. The fire image recognition model was established by using MPCANet. Through the MPCA algorithm, the learning filter was used as the convolution kernel of deep learning network convolution layer, and the feature extraction of high dimensional images of tensor objects was taken, and candle images and fireworks images were taken as interference. Compared with other fire image recognition methods, the recognition accuracy of the proposed image recognition method reaches 97.5%, false alarm rate of 1.5%, missing rate of 1%. Experiments results show that this method could effectively solve the problems of fire image recognition.
2018, 47(2): 203007.
doi: 10.3788/IRLA201847.0203007
Action recognition from natural scene was affected by complex illumination conditions and cluttered backgrounds. There was a growing interest in solving these problems by using 3D skeleton data. Firstly, considering the spatio-temporal features of human actions, a spatio-temporal fusion deep learning network for action recognition was proposed; Secondly, view angle invariant character was constructed based on geometric features of the skeletons. Local spatial character was extracted by short-time CNN networks. A spatio-LSTM network was used to learn the relation between joints of a skeleton frame. Temporal LSTM was used to learn spatio-temporal relation between skeleton sequences. Lastly, NTU RGB+D datasets were used to evaluate this network, the network proposed achieved the state-of-the-art performance for 3D human action analysis. Experimental results show that this network has strong robustness for view invariant sequences.
2018, 47(2): 203008.
doi: 10.3788/IRLA201847.0203008
Learning rich representations efficiently plays an important role in RGB-D object recognition task, which is crucial to achieve high generalization performance. For the long training time of convolutional neural networks, a Hybrid Convolutional Auto-Encoder Extreme Learning Machine Structure (HCAE-ELM) was put forward which included Convolutional Neural Network (CNN) and Auto-Encoder Extreme Learning Machine (AE-ELM), which combined the power of CNN and fast training of AE-ELM. It used convolution layers and pooling layers to effectively abstract lower level features from RGB and depth images separately. And then, the shared layer was developed by combining these features from each modality and fed to an AE-ELM for higher level features. The final abstracted features were fed to an ELM classifier, which led to better generalization performance with faster learning speed. The performance of HCAE-ELM was evaluated on RGB-D object dataset. Experimental results show that the proposed method achieves better testing accuracy with significantly shorter training time in comparison with deep learning methods and other ELM methods.
2018, 47(2): 203009.
doi: 10.3788/IRLA201847.0203009
Fatigue driving is the main cause or reason for traffic accidents, which has a huge influence on social safety. Considering the fact that light change and glasses could significantly increase the difficulty to monitor human eyes, fatigue detection was still an unsolved problem. A new driver fatigue method based on morphology infrared features and deep learning were proposed. Using 850 nm infrared light source, the facial image was obtained. Human faces and landmarks which indicated the area of eyes were located by Convolution Neural Network (CNN) with morphology features in infrared image. In the next step, a filter module which measured head displacement was added, aiming at reducing the impact of posture change. In the following, the collected facial states were transformed into sequential data. Finally, the sequential data was passed to the Long Short Term Memory (LSTM) network to detect fatigue state by analyzing the sequential correlations. Experimental results show that the accuracy of the fatigue detection algorithm can reach 94.48% with an average detection time of 65.64 ms.
2018, 47(2): 204001.
doi: 10.3788/IRLA201847.0204001
Subjected to the material quality and the device fabrication techniques, the IRFPAs usually suffer from bad pixels and stripe noise. The accuracy of the bad pixel detection is always influenced by the stripe noises, which should be optimized for more precise imaging processing. A preliminary clean image was accomplished without the influence of stripe noises by reconstructed double density double tree complex wavelet coefficients with different weights. After that, the image preprocessing was utilized to detect the bad pixels with 3 criterion. Finally, this method was testified by a image processing from short-wave IRFPA. The bad pixels detection in the infrared image is much more accurate in spite of the stripe noise.
2018, 47(2): 204002.
doi: 10.3788/IRLA201847.0204002
When detecting a target over the diurnal cycle, due to the large temperature variation of targets and background objects, the infrared image contrast between target and background in a scene would probably approach to zero, by which the target would be blended into the background in thermal infrared image, causing the blindness of thermal infrared system, lowering the detection accuracy and even making the thermal sensor lose the target. Concerning this issue, based on the brief theoretical analysis on how the thermal crossover over the diurnal cycle influenced a conventional thermal infrared system, a mid-infrared (3-5 m) multi-spectral method using different infrared spectral characteristics of different targets to solve the issue was presented. Furthermore, the prototype design and how multispectral technology was employed to help solve the thermal crossover detection problem were also described. The experimental results show that the multi-spectral infrared imaging system can increase the contrast of the detected images by 18% compared to the MWIR broadband image and effectively solve the failure of the conventional infrared sensor during the diurnal cycle, which is of great significance for infrared surveillance applications.
2018, 47(2): 204003.
doi: 10.3788/IRLA201847.0204003
The detectability of See-To-Ground (STG) wavebands of the infrared warning satellite computational models were constructed, and the possible wavebands range and the role of this STG waveband in missile early warning system were studied by simulation on the basis of those models. Firstly, a series of models for the infrared radiation model of solid and liquid rocket plume, the atmospheric background radiation, the point target radiation flux apparent contrast spectrum, the system noise and the lowest detection height were established. Secondly, the apparent contrast spectrum of solid and liquid rockets plume were analyzed and the STG band was deemed to be 2.86-3.0 m. Thirdly, the detection ability was preliminarily investigated based on the analysis of the rule by which the apparent contrast spectrum of solid and liquid rockets plume varied according to the height. Finally, the detectability of SBIRS-GEO warning satellite at STG band was studied thoroughly, based on the analysis of the first detection height under different condition. The research shows that the STG band possesses superior early warning ability to solid rockets, and weaker early warning ability to liquid rockets.
Aiming at the problem of thermal barrier effect, transient temperature rise of sphere dome was calculated by fluid-solid coupling method, and then the structure of dome was improved. Based on wind tunnel and rocket sled tests, as well as analysis of the potential of vorticity interaction in entropy layer, the critical Reynolds number based on the free stream parameters and the diameter of dome which could be used to judge the flowing state over sphere dome was proposed as 2.7106. Compared with the experimental data, the laminar heat transfer formula of Van Driest based on isentropic relations was found to be applicable for the calculation of heat transfer for the most area of hemisphere dome. Based on Van Driest's formula and finite element method, the heat transfer of dome was calculated to obtain the transient temperature field of sphere dome. The results show that temperature rise of dome is relatively fast and the trajectory should be well designed to avoid thermal barrier effect. For the requirement of high speed long-endurance system, a composite optical dome structure was designed based on porous ceramic, which could extend the flight time significantly and compared with side-window system, the aero-optic effects caused by the new composite dome was small. The new composite optical dome is suitable for long wave infrared optical systems. Thermal shock stress in composite optical dome is smaller, and because of conformal shape, the aerodynamic drag of composite optical dome is smaller also. Then the composite optical dome can be flexibly applied under the flight conditions of high mach number and low altitude.
2018, 47(2): 204005.
doi: 10.3788/IRLA201847.0204005
Several kinds of artificial biological materials have the characteristics of low production cost, low mass density, controllable morphology, safety and environmental friendliness, and better extinction ability in the visible and infrared bands, which can be used as new extinction materials to make up for the shortage of inorganic extinction materials. With the increasing demands for new biological extinction materials in military and civilian fields, the flocculent biological particles in the biological extinction material are equivalent to bullet rosette particles. According to Newton's second law and gradient transport theory, the sedimentation and diffusion models of biological extinction material were established. The structures of flocculent biological particles and wind speed and atmospheric stability' influences on aerodynamic characteristics of biological extinction materials were discussed. The results indicate that the structures of flocculent biological particles are controlled by the number of branches, as well as their length and radius. Under given conditions, the settling rate of flocculent biological particle decreased by 50% compared with that of equal volume sphere particle. And the max covering area of smoke can reach more than 20 m2. The model provides a theoretical basis for the further development and use of biological extinction materials.
2018, 47(2): 206001.
doi: 10.3788/IRLA201847.0206001
Aiming at improving the damage probability of ground attack rocket effectively, the laser circular-viewing detection system was adopted to capture the target credibly. Computational formulas for the laser echo and the minimal detectable optical power were proposed to establish the model of the missile-target encounter based on the terminal trajectory characteristics of ground attack rocket. The change regulation of laser repetition frequency and motor scanning speed on the target capture probabilities using different detection systems were analyzed by using Monte-Carlo method to select the optimal repetition frequency and scanning speed. And the effects of rocket projectile velocity and hit accuracy on target capture probabilities were discussed. The experimental results show that the damage probability of single guided rocket projectile is increased effectively by using the laser circular-viewing detection system. Meanwhile the laser repetition frequency is 5 kHz, and scanning speed is 10 000 r/min. Simulated outcome provides the basis for the application of the laser circular-viewing detection system on guided rocket projectile.
2018, 47(2): 206002.
doi: 10.3788/IRLA201847.0206002
Based on laser speed measurement, a method for high-accuracy speed measurement was developed by using femtosecond optical frequency comb based on the phase signal processing. The low speed measurement limit is m/s level theoretically. The principle of phase changing was analyzed in detail, and a comprehensive model was established and simulated with MATLAB. The phase information of simulated measurement signals were processed by fast Fourier transform, and the speeds of m/s, mm/s and m/s were measured respectively. Ideally, the measurement error was below 1 m/s. In addition, arbitrary movement targets can be measured through the model. The movement state of target was restored with this method. The experiment was conducted to measure the speed of mm/s, and the relative errors were all within 4%. The results show that high-accuracy speed measurement can be achieved by using this method.
2018, 47(2): 206003.
doi: 10.3788/IRLA201847.0206003
Aiming at the problem that the target signal may be submerged in the water scattering signal when the detector is near the ship wake, as well as the difficulty of detecting the low density and small bubbles, based on our previous researches, a method was proposed by comparing and analyzing the differences of the polarization effect of scattering characteristics when laser irradiated the micro-bubbles, water, we could distinguish bubble signals, as well as by measuring the Stokes parameters of scattered light, this can fully extract the information of ship wake field's features. The scattering of the Stokes scattering matrix under different radius and scattering angles was calculated and compared with the scattering characteristics of water molecules. The results show that:(1) Stokes scattering characteristics of wake waves with different sizes have similarities and differences. (2) The scattering parameters of microbubbles are more obvious than those of water molecules. There is no oscillatory phenomenon in the scattering parameters of water molecules and the scattering characteristics are stable. Using the Stokes parameter to analyze the characteristics of ship wake, the information volume is more comprehensive, and the light scattering characteristics are obviously different, which appear to be useful for the analysis and detection of wake signal.
2018, 47(2): 206004.
doi: 10.3788/IRLA201847.0206004
The atmospheric refraction delay caused by the uneven distribution of the refractive index of the atmosphere, is one of the main sources of the ranging error for a space-borne laser altimeter and mainly influenced by the surface pressure. At present, the method of calculating surface pressure is based on NCEP meteorological data and the time-space interpolation method derived by GLAS research team. For the GLAS system, the correction precision is sufficient (1 cm) for its core observing target in the relatively flat ice-sheet region; however, the correction precision is low for a complex land target. Based on the observation data of China domestic meteorological stations, a modified inverse-distance weighted interpolation algorithm was derived to improve the correction precision of the atmospheric refraction delay; then, the correction precision was compared with traditional GLAS method. In the highland and high latitudes area, the correction error was reduced from more than 2 cm to less than 0.5 cm using the new interpolation algorithm with domestic meteorological data, which was equivalent to improve the overall accuracy of 10% compared to the whole 15 cm measurement accuracy of GLAS system. This research will be reference to improve the measurement accuracy for future domestic space-borne laser altimeter.
2018, 47(2): 206005.
doi: 10.3788/IRLA201847.0206005
The technology of laser feedback has ultra high sensitivity and great advantages in vibration measurement. Based on laser feedback theory and technologies, method of remote vibration measurement with laser feedback technologies was studied and presented. The remote vibration measurement system based on microchip solid-state laser feedback was built, the structure and working principles of each parts of the system were studied in detail. The vibration signals of different frequencies in experiments were measured and restored. The working distance of vibration measurement was improved to 25 meters, realizing non-contact measurement of remote vibration. The application of laser feedback technology was expanded. The experimental system has superior performance and large range of amplitudes and frequencies in vibration measurement, and it can be applied in various occasions and can meet various target measurement requirements.
2018, 47(2): 230001.
doi: 10.3788/IRLA201847.0230001
Doppler wind lidar with its high resolution, high precision, large detection range, the ability to provide three-dimensional wind field information, has attracted the attention of multinational scholars, and put a lot of manpower and material resources to carry out research. The Doppler wind lidar was designed to obtain single longitudinal mode, narrow linewidth and high power laser output by seed injection. The slow drift of the center frequency of the laser, ambient noise, laser rod temperature change or the vibration disturbance may cause the failure of seed injection. Then the laser spectrum was transformed from a single longitudinal mode output to a multiple longitudinal mode output. The linewidth of single longitudinal mode laser output was about 200 MHz. However, the multi-longitudinal mode laser output had a wide linewidth,which led to widened Rayleigh backscattering spectrum and then great speed errors in the speed inversion condition. The pulse screening circuit was developed to filter the multi longitudinal mode pulse during the data acquisition, which can effectively reduce the error of wind speed inversion and improve the precision of wind speed measurement.
2018, 47(2): 230002.
doi: 10.3788/IRLA201847.0230002
The precision of wind filed retrieved from coherent wind lidar scanning measurement mode using VAD method could be substantially low without quality control. By investigating error sources in the scanning measurement mode, based on least square VAD fitting algorithm, a stepwise regression procedure for quality control of the data involved in wind regression was presented. The statistical parameters adopted for the quality control process included SNR, data residual, scanning interval effective data and the data efficiency. Based on the technique proposed, the quality control strategy was designed and verified by comparing the Lidar retrieved 2 955 pair of 10-minuts averaged wind data with those derived from the wind cup data. The result shows that the root mean square deviation (RMSD) of the wind speed has been reduced for approximately 44% from 0.97 m/s to 0.54 m/s, while the RMSD of wind direction has been reduced for approximately 26% from 7.47 to 5.55.
2018, 47(2): 230003.
doi: 10.3788/IRLA201847.0230003
The staring imaging lidar, whose indexes selection influences the load of spacecrafts seriously, is an important method for the pose measurement of non-cooperation target in space. Selecting the indexes of lidar reasonable is significant, opposite to the traditional method by experience, making it not only satisfy the demands of space mission, but also effectively reduce the load of spacecrafts and energy consumption. Macro model for compatibility between indexes of on-orbit lidar and attribute of target was proposed to provide a theoretical basis for indexes selection of lidar. Four indexes included lidar output power, lidar resolution, lidar field angle and lidar ranging precision. The processes were as follows. The initial values of the lidar resolution and field angle can be calculated by the scale of the target and distance between lidar and target. The transmitting power can be acquisited by iterating the maximum measurement distance. The reasonable resolution and field angle can be measured by iterating the second order parameters of the target surface depth co-occurrence matrix. The ranging precision of the lidar can be acquired by iterating the result of the point clouds registration algorithm. The experimental result of lidar indexes selection shows that proposed model can satisfy the requirement for the lidar parameters optimal selection.
2018, 47(2): 230004.
doi: 10.3788/IRLA201847.0230004
Streak array detecting lidar has advantages of high ranging accuracy, long detecting distance, wide ranging gate and high data acquiring rate, which is widely used in topographic mapping, coastal zone monitoring, urban 3D remodeling, forest ecological research etc. The traditional method on signal discrimination has some limitations in the depth extraction of echo signal, which affects the capability of depth distinguish and target recognition of streak array detecting lidar. For this problem, an iteratively weighted centroid algorithm was introduced while dealing the streak images and the unique advantages on centroid location of the algorithm were discussed. The key parameters of the algorithm were determined according to signal distribution character of streak array detecting lidar. By using this algorithm, the range image with clear boundary feature of a target was obtained at the distance of 1.4 km. The boundary blurring effect was effectively suppressed in the depth extraction process and the resolution ability of the system was improved. Compared with traditional centroid algorithm, the ranging accuracy was improved by 17%.
2018, 47(2): 225001.
doi: 10.3788/IRLA201847.0225001
Considering the terahertz near-field scattering characteristics, a high-frequency terahertz radar cross section (RCS) measurement system was presented based on a CO2 laser pumped terahertz laser and an independent double-deck rotation platform. The reliability of the system was verified by utilizing stainless steel smooth spheres as the standard calibration targets. The error between measurement result and the theoretical value was less than 3 dBsm and the signal-to-noise ratio is better than 24 dB. For the first time, RCS measurement of metal plates was performed with different materials and coatings, and metal cones with different bottom diameter, at a high frequency of 3.11 THz. The RCS difference of pure, anodized, painted aviation aluminum and P304 stainless steel plates, was demonstrated. The RCS of cones with different bottom diameters made of pure aviation aluminum plate was also compared. This work can lay the foundation for further study of the complex target RCS research in the terahertz band.
2018, 47(2): 225002.
doi: 10.3788/IRLA201847.0225002
The modified polypropylene(PP) material has been widely used in dashboard, bumper and other auto parts because of its characteristics of flame retardant, high impact resistance, etc. It is a necessary means to improve the quality of auto parts by using the nondestructive testing technology to detect the modified PP material. A transmission terahertz time-domain spectroscopy(THz-TDS) imaging system and a reflective THz-TDS imaging system were built. The optical parameters of modified PP materials were extracted by adopting the transmission THz-TDS system. The refractive index, one of the optical parameters in terahertz band can be solved through the mathematics model, its value was 1.53. A kind of modified PP material sample with flat bottom holes was designed. The reflective THz-TDS imaging system was adopted for measuring the sample, the deconvolution method was used to filter the time domain THz signal, which largely improved the signal-to-noise ratio. What's more, THz-TDS tomography imaging technology was proposed based on the time of flight. Combined with the measured refractive index, the 3D structure of the modified PP material was reconstructed with the time of flight tomography imaging technique, the thickness accuracy is 0.01 mm.
2018, 47(2): 222001.
doi: 10.3788/IRLA201847.0222001
An adaptive multiple-input multiple-output orthogonal frequency division multiplexing (MIMOOFDM) visible light communication(VLC) system based on space-time block coding (STBC) modulation was proposed in the paper, which can overcome the MIMO channel correlation and achieve reliable communications. Meanwhile, power-and-bit allocation(PBA) was combined with OFDM in order to adapt to the channel and improve the spectral efficiency. Then a 22 MIMO-OFDM VLC transmission over a distance of 80 cm was demonstrated, where the measured bit error rates were all below 7% of the pre-forward-error-correction threshold of 3.810-3. Experimental results show that the STBC MIMO-OFDM system is robust to MIMO channel correlation, where the data rates can be greatly improved by applying the PBA.
2018, 47(2): 222002.
doi: 10.3788/IRLA201847.0222002
The sensing performance of micro-ring biosensor on Silicon-On-Insulator (SOI) was studied through simulation. The sensitivity of 38.71 nm/RIU, detection limit of 1.810-3 RIU and Q factor of 2.22104 were achieved. The effects of noise on the sensor, including the light source noise and temperature noise, were then investigated. A dual micro-ring differential sensor with reference and probe channels was proposed, and the wavelength drift induced by the noises can be deducted to reduce the influence of noise by this differential operation. The relative error of the refractive index change was reduced by 15.85% through the calculation and numerical simulation, so the micro-ring differential biosensor can effectively reduce the influence of noise to enhance the sensing performance of the micro-ring biosensor.
2018, 47(2): 222003.
doi: 10.3788/IRLA201847.0222003
A linear optical current transducer (OCT) based on radial polarization demodulation was proposed. A magnetic concentration ring with an air gap was selected as the sensing head, in which two magnetizer plates were used to lead out the magnetic field of air gap, and the straight optical path passed through the air gap by light apertures. The reflecting prism was not needed so that the reflection-induced retardation was eliminated, which helped to improve the stability of the transducer. A radially polarized grating was employed to convert the Faraday rotation angle into the rotation of a circular facula, which was detected by a four-quadrant detector, and then the measured current could be obtained after signal processing. The theoretical analysis and experimental results show that this new OCT has a measurement range of 43 in Faraday rotation angle and the advantage of optical power independent, meeting the national standard of 0.2S accuracy class.
2018, 47(2): 220001.
doi: 10.3788/IRLA201847.0220001
In order to realize dynamic interference measurement of Fizeau interferometer, a dynamic Fizeau interferometer using low-coherence light source was proposed. A diode-pumped solid-state laser with a central wavelength of 638 nm and a bandwidth of 0.1 nm as the light source was used to produce a pair of orthogonal linearly polarization lights with low-coherence, combined with a polarization delay device. Then a pair of coherent beams with zero optical path difference (OPD) were acquired through adjusting OPD of the linearly polarization lights to match cavity length of Fizeau interferometer. Four interferograms with /2 phase shift in a phase-shifting sequence were obtained by a pixelated polarizer-based camera, by using the 4-bucket algorithm, the surface shape of optical elements was derived. An optical intensity normalization algorithm was adopted to suppress the phase error, which was caused by the intensity inconsistency due to polarization error, and had the same spatial frequency as the fringes. The relationship between interferogram contrast and ratio of s-polarized light and p-polarized light was deduced through Jones vector and Jones matrix. Besides, the effect of azimuth error of the quarter wave plate on the final recovery wave surface was analyzed. An optical flat was measured by this interferometer and Zygo GPI XP interferometer, the difference of root-mean-square and peak-to-valley were 0.024 and 0.026 respectively.
2018, 47(2): 220002.
doi: 10.3788/IRLA201847.0220002
In order to obtain wide-band and high-resolution monochromatic light, the wavelength of the imaging spectrometer was calibrated, and a scanning monochromator was designed. The grating scanning system was driven by worm and worm wheel mechanism, and an off-axis installation method was designed to install the worm gear, thus solving the problem of the traditional installation method, such as the decrease of the effective aperture of the grating, and the increase of the stray light. The monochromator optics system used the horizontal Czerny-Turner structure. It used three gratings to achieve 280-2 240 nm wide band output, which achieved the high diffraction efficiency and guaranteed the spectral resolution in the whole wavelength range. The output wavelength and the motor step number was non-linear in the process of worm gear scanning, so the monochromator system was calibrated by different mathematical models. The final experiment and the measurement proved that the spectral resolution of the monochromator was better than 0.1, 0.2, 0.4 nm at 280-560 nm, 560-1 120 nm, 1 120-2 240 nm respectively. Simultaneously the wavelength repeatability reached to 0.094, 0.186, 0.372 nm, and the precision reached to 0.096, 0.191, 0.382 nm. The monochromator achieves the design goals, and meets the requirements of the wavelength calibration of the imaging spectrometer.
2018, 47(2): 220003.
doi: 10.3788/IRLA201847.0220003
Micro-distance spectra splitting is a key technology for compact dual-band imaging system in infrared field. Utilizing the Fresnel diffraction theory and angular spectrum theory, the diffractive properties of dual-wavelength and wide angle light passing through stepped phase grating in fractional Talbot distance were theoretically investigated. A typical scheme of imaging structure based on dual-array detection pixels was set up, where two light beams with different wavelength incident after passing through the stepped phase grating. The numerical simulation shows the splitting results of incident 4 m and 4.5 m wavelength light with-30 to 30 angle, including the distribution of luminous intensity, energy and signal-to-noise ratio on dual-array detection pixels. Our results are beneficial not only for designing the compact dual-band optical imaging structure, but also for building up precision spectral separating component.
The confidence level of rotation angle error test result was reduced by nonparallelism between photoelectric shaft encoder axis and polyhedron one. In order to decrease rotation angle test error, the rotation angle test error caused by nonparallelism between photoelectric shaft encoder axis and polyhedron one was controlled within 1/3-1/5 of photoelectric shaft encoder angle error. The mathematical models of rotation angle test error and Y bias introduced by nonparallelism between photoelectric shaft encoder axis and polyhedron one were established. The simulation results showed that rotation angle test error and Y bias had periodic change with the angle increase, the cycle were and 2 respectively. When the tilt direction of polyhedron axis was fixed, the bigger angle between encoder axis and polyhedron one, the bigger peak values of angle test error and Y bias, when the angle between photoelectric shaft encoder axis and polyhedron one was fixed, the tilt direction of polyhedron axis only changed the phases of rotation angle test error curve and Y bias one and not changed the shapes of the curves. According to the polyhedron-autocollimator method, the established mathematical models were verified by experiments. Experimental results indicate that the test results and the mathematical models are in good conformity. In the actual test, the rotation angle test error is pretested, the curve is drawn and fitted by the least squares method, the parallelism and tilt direction are obtained, the parallelism is adjusted according to tilt direction until error peak satisfies test requirement.
2018, 47(2): 217002.
doi: 10.3788/IRLA201847.0217002
Aiming at the problem of long time and slow positioning speed when the digital zenith camera was in precise leveling state for astronomical positioning, based on the principle of positioning the digital zenith camera, the correction model of the inclination angle was established by the direction cosine matrix transformation principle, the positioning method after the correction of the tilt angle was analyzed and deduced, and the two kinds of localization algorithms of the tangent plane and the spherical triangle of the digital zenith were improved, achieved the positioning of equipment without fine leveling, speeded up the positioning speed of the purpose. The accuracy of the two types of positioning algorithms in the tilted state was compared by experiment. Experiments show that the positioning accuracy of the spherical triangle method is relatively high, and the calculation accuracy of the latitude and longitude can reach 0.5 or less.
2018, 47(2): 217003.
doi: 10.3788/IRLA201847.0217003
To analyze test errors induced from system noise of swing arm profilometer (SAP) testing large aperture off-axis aspherics with different scan arcs, based on the Monte Carlo method, the simulation test model was built with different scan patterns from 8 arcs to 120 arcs. To evaluate the test sensitivity, test errors induced from system noise was fitted by Zernike polynomial. The low order test errors were reduced rapidly as scan arcs increased from 8 arcs to 39 arcs, and reduced slowly from 40 arcs to 70 arcs, and kept almost the same from 71 arcs to 120 arcs. As an experiment, a 1 500 mm off-axis aspheric mirror was tested by SAP scanning in 36 arcs, 72 arcs and 96 arcs respectively. The surface test error of 36 arcs are much larger with result of 7.73 m in Peak-Valley (PV) and 0.68 m in RootMean-Square (RMS). The surface test error of 72 arcs and 96 arcs are much close with result of 5.755 m PV, 0.568 m RMS and 5.612 m PV, 0.569 m RMS respectively. So the simulation results were verified and provided a guide for the scan line selection for SAP testing large off-axis aspherics.
2018, 47(2): 217004.
doi: 10.3788/IRLA201847.0217004
Tissue models play an important role in the study of the imaging mechanism of biological tissues and in the research and improvement of optical disease diagnosis technology. However, there are not only spherical scatterers in epithelial cells, but also ellipsoidal scatterers generally. In this paper, there are few studies of epithelial tissue models containing ellipsoidal scatterers so far. Sphere-rotation ellipsoid scattering model was established, and the effect of model parameters on forward scattering intensity distributions and Mueller matrix polar decomposition parameters were analyzed by Monte Carlo simulation method. The results show that the increase of scattering coefficient, the increase of volume and axial ratio of rotation ellipsoids all would lead to a higher depolarization power. Moreover, the influence of scattering coefficient of rotation ellipsoids on depolarization is most significant. The ellipsoidal scatterers with an anisotropic distribution contribute to retardance, which has positive correction with the scattering coefficient and axial ratio of ellipsoids and negative correction with the volume of ellipsoids. This study can give some clues to the development of precancer diagnosis technology.
2018, 47(2): 226001.
doi: 10.3788/IRLA201847.0226001
Curvature filtering-empirical mode decomposition preprocessing was used to detect and extract human target features. It reduced the computational complexity of image decomposition, enhanced edge and texture features simultaneously, and improved feature differentiation. Its performance was in the following areas:(1) In the first layer of empirical mode decomposition, the continuous smooth surface of the original image was mapped by curvature filtering plane to form the envelope surface and the mean surface. The first layer extracted texture features, and the following layers highlighted structure features. (2) Matching regions had sharp edges from adjacent layers which varied from low-resolution to highresolution, easy to extract the target contour candidate regions. (3) Decomposing the texture features in the first floor to screen the background, matching the foreground feature areas in the adjacent layers to form the trajectory map of the human body posture, it made easy judging the human body posture and behavior. When applying it to human behavior recognition experiments, and comparing it with the ground truth of the human behavior database, the accuracy of the contour extraction and the recall rate of the UIUC sample data were all over 90%. And experiments verify the preprocessing method is effective in human pose recognition and behavior recognition.
2018, 47(2): 226002.
doi: 10.3788/IRLA201847.0226002
Hyperpectral unmixing is a difficult problem in academia. Sparse hyperspectral unmixing uses priori spectral library, aiming at finding several pure spectral signatures to express hyperspectral images and computing corresponding abundance fractions. This is NP-hard to solve. Convex relaxation for L0 norm as L1 norm is a common approach to solve the sparse unmixing problem, but only approximation results can be achieved. A Pareto optimization based sparse unmixing algorithm was proposed(ParetoSU). ParetoSU firstly transformed sparse unmixing to a bi-objective optimization problem. One of the two objectives was the modelling error and the other one was the sparsity of endmembers. ParetoSU can solve the sparse unmixing problem without any approximation of L0 norm. At last, synthetic data were used to test the performance of ParetoSU.
