Application of Object-Oriented Median Absolute Deviation Method to Building Extraction
Gong Xunqiang, Liu Xinglei, Lu Tieding, and Chen Zhigao
Buildings are extremely important artificial feature objects. Extracting buildings can provide technical support for urban planning, population estimation, and landscape analysis. Object-oriented classification is an effective method for extracting ground objects and has been widely used in the extraction of building information. The object-oriented morphological building index method has good practicability, but the effect of extracting sparse buildings still needs to be improved. To solve this problem, the median absolute deviation is applied to the object-oriented building extraction, and the two situations of dense and sparse buildings are analyzed. Precision, recall, and F1 score are used to evaluate the extraction results. Experimental results show that the object-oriented median absolute deviation method extracts sparse buildings significantly better than the object-oriented classification and object-oriented morphological building index methods.
  • Jun. 22, 2021
  • Laser & Optoelectronics Progress
  • Vol.58 Issue, 12 1212005 (2021)
  • DOI:10.3788/LOP202158.1212005
Strain Field Calculation by 3D Digital Image Correlation Method Based on Subset Projection and Savitzky-Golay Filter
Gu Jun
The digital image correlation (DIC) method is a non-contact, full-field deformation measurement method. The three-dimensional DIC breaks through the requirement that the surface of the test piece is planar in the two-dimensional DIC system, and obtains the three-dimensional surface displacement field data of the test piece through a stereo or multi-view camera system. First, a local subset is used to fit the least square plane, and the discrete points on the three-dimensional spatial curved surface are projected into the two-dimensional plane. Then, the Savitzky-Golay filter is used to solve the local strain tensor. Finally, the characterization of the strain field of a free-form test piece is realized, especially the errors of the small-plane projection method are investigated. The results show that the calculation error of the strain field caused by the projection process is much smaller than the transmission of the systematic error component contained in the displacement field data. When the strain subset size M=5, the strain field error is about 30 με. When M=10, the strain field error is about 10 με.
  • Jun. 22, 2021
  • Laser & Optoelectronics Progress
  • Vol.58 Issue, 12 1212004 (2021)
  • DOI:10.3788/LOP202158.1212004
Single Pixel Detection Theory of Flat Surface Reflectivity Anomaly
Ouyang Haoyi, Chen Wanjun, Li Hai, and Yang Chuping
To detect the abnormal quality of smooth surfaces under standardized quality control, a single-pixel detection theory about the abnormal reflectivity of a flat surface is proposed. The proposed method uses a single-pixel detector and only needs to project a single frame of structured light (illumination design). First, the radiation flux distribution of a single-pixel detection about the abnormal reflectivity of a flat surface is derived. It is shown that under uniform illumination conditions, the spatial distribution of the radiation flux to the detector is nonuniform. Thus, a special illumination design can achieve the uniform radiation flux distribution and convert the abnormal reflectivity distribution of a flat surface into the cumulative reflectivity anomaly (or total radiation flux anomaly). In the experiment, the corresponding detection device is designed. The radiation flux distribution on the flat surface to the detector under uniform illumination and the illumination design for achieving uniform radiation flux distribution are numerically calculated. They are consistent with the actual results. Under the illumination condition of uniform radiation flux distribution, the total radiation flux of seven types of qualified ceramic tiles and the abnormal total radiation flux caused by two kinds of surface defect—cracks and scratches were investigated. The results showed that the two surface defects lead to significant changes in the total radiation flux. The effectiveness of the theory and feasibility of the technology are preliminarily verified.
  • Jun. 22, 2021
  • Laser & Optoelectronics Progress
  • Vol.58 Issue, 12 1212003 (2021)
  • DOI:10.3788/LOP202158.1212003
Laser Dazzler with Automatic Focal Length Adjustment for Pedestrians
Cui Hongsheng, and Jiang Tao
Police laser blinding devices are primarily used in the field of security. A human target will lose vision temporarily when exposed to the green laser emitted by the device because the human eye is sensitive to the 532-nm green laser. The laser light must cover the target human body as much as possible to improve utilization of the laser. This subject obtains video from the camera and detects pedestrian targets through a video processing system based on digital signal processing (DSP) using the Gentle Adaboost algorithm. The detection results include the size and location of the pedestrian targets. According to the size of the pedestrian, a single chip microcomputer control circuit controls the laser lens to focus on pedestrian targets automatically to accommodate pedestrian targets.
  • Jun. 22, 2021
  • Laser & Optoelectronics Progress
  • Vol.58 Issue, 12 1212002 (2021)
  • DOI:10.3788/LOP202158.1212002
Three-Dimensional Shape Measurement Based on Hybrid Dual-Frequency Fringe Projection
Liu Lu, Xi Dongdong, Chen Zhijian, Cheng Lei, and Wang Yuwei
The dual-frequency fringe projection is an important method for three-dimensional (3D) shape measurement. In the actual measurement process, how to balance the speed and accuracy has a remarkable impact on practical applications. This paper introduces a hybrid dual frequency algorithm, which needs to project four fringe images, including three high frequency fringe images and one low frequency fringe image. First, the background intensity and truncation phase of high frequency fringes are calculated based on three-step phase shift. Second, the background intensity is subtracted from the low-frequency fringes, and the low-frequency truncated phase is extracted by Hilbert transform. The low frequency truncation phase is unwrapped by the geometric constraint method, and the low frequency truncated phase is used to unwrap the high-frequency truncated phase. In order to verify the effectiveness of the proposed method, simulation and experimental studies are carried out respectively. Experimental results show that the proposed method can effectively restore the three-dimensional shape of the measured object.
  • Jun. 22, 2021
  • Laser & Optoelectronics Progress
  • Vol.58 Issue, 12 1212001 (2021)
  • DOI:10.3788/LOP202158.1212001
Application of Genetic Algorithm in Particle Size Measurement Technology Based on Forward Light Scattering
DAI Jun, and SHEN Jianqi
In the technology of particle size measurement based on forward light scattering, the inversion algorithm is one of the key issues affecting the accuracy of the measurement results. As a relatively new adaptive global optimization algorithm, genetic algorithm parameters optimization and objective function design determine the accuracy of inversion results of particle size distribution. In this paper, the setting of relevant parameters of genetic algorithm is discussed, and the numerical simulation results show that the algorithm is effective in the inversion of the particle size distribution of polydisperse particle system. The measurement and inversion calculation of polystyrene standard particles size distribution were carried out. The results prove that the optimized genetic algorithm which the genetic iteration is 3 200, population scale is 50, crossover operator is 0.7, mutation operator is 0.002 and the weight coefficient of fairing factor is 1.5×10-5 has good stability and noise immunity, and can effectively reconstruct the particle size distributions.
  • Jun. 22, 2021
  • Acta Photonica Sinica
  • Vol.50 Issue, 5 213 (2021)
  • DOI:10.3788/gzxb20215005.0512002
Quality Assessment Method of Speckle Patterns Based on Correlation Coefficient of Adjacent Subsets
LI Xiangbo, GONG Jun, HU Dan, SONG Jianwen, and LIU Kai
By analyzing the relationship between the correlation coefficient of adjacent subsets and the quality of speckle patterns, a new speckle quality assessment parameter with a fixed range called mean adjacent zero mean normalized cross correlation is proposed, which can be used to assess the quality of single speckle pattern. By comparing the proposed mean adjacent zero mean normalized cross correlation, Shannon entropy and mean intensity gratitude, the stability of the proposed method is founded. The experiment results show that as the mean adjacent zero mean normalized cross correlation increases, the mean error and standard deviation in the displacement measurement decreases, and the quality of the simulated speckle pattern become better. When the maximum mean error of the speckle pattern in the displacement experiment is 0.01 pixel, mean adjacent zero mean normalized cross correlation is stable at around 0.90, while the other parameters vary greatly. These results prove that mean adjacent zero mean normalized cross correlation is effective and more suitable for quality assessment of single speckle pattern compared with other quality assessment parameters.
  • Jun. 22, 2021
  • Acta Photonica Sinica
  • Vol.50 Issue, 5 204 (2021)
  • DOI:10.3788/gzxb20215005.0512001
Polarization Parameter Validation of Directional Polarimetric Camera Based on Cloud Polarization
TU Bihai, YAO Pingping, LI Shu, TI Rufang, MENG Binghuan, HUANG Chan, and HONG Jin
According to the simulation results of cloud polarization radiative transfer, the water cloud with polarization rainbow characteristics is selected as the experimental target to test the space environment adaptability of key polarization parameters. On the basis of the characteristics of polarization imager directional polarimetric camera, the large area water cloud is identified by cloud phase state. The relative transmittance and polarizability parameters are verified according to the polarization radiation model of the instrument by using the measurement data of different polarizing directions of water cloud. The experimental results show that the change of relative transmittance is less than 0.2%, the variation of polarizing degree for optical lens is less than 0.01, and the instrument state is stable. At the same time, the ability of polarization imager to identify cloud phase state by cloud polarization characteristics is verified. The methods and test results provide reference for on orbit detection and calibration of polarization imager, and provide basis for inversion of cloud products using multi angle polarization data.
  • Jun. 22, 2021
  • Acta Photonica Sinica
  • Vol.50 Issue, 5 223 (2021)
  • DOI:10.3788/gzxb20215005.0512003
Comparison on Noise Resistance of Temporal Phase Unwrapping Algorithms
Shi Lili, Liu Weiguo, Tian Ailing, and Liu Bingcai
Based on the three-dimensional (3D) measurement technology of grating fringe projection, phase principal values are extracted using the phase shift method. The accuracy of the phase principal value expansion is the key factor in determining the accuracy of the entire 3D measurement system. Temporal phase unwrapping algorithms can realize the phase unwrapping of isolated and discontinuous objects; however, owing to the influence of sensors and environmental noise, the phase unwrapping result has large errors. To select a good antinoise algorithm, herein, several temporal phase unwrapping algorithms were compared using simulations and experiments. Results show that the multi-frequency (hierarchical) phase unwrapping and negative exponential fitting methods exhibit good antinoise performance and high phase resolution accuracy. Alternatively, the multi-wavelength (heterodyne) method and multi-wavelength (heterodyne) method based on the fringe location exhibit a poor antinoise performance and large phase errors. The findings of this study can provide guidance to researchers in selecting phase decomposition algorithms, and they have important application values.
  • Jun. 21, 2021
  • Laser & Optoelectronics Progress
  • Vol.58 Issue, 12 1212006 (2021)
  • DOI:10.3788/LOP202158.1212006
Frequency Modulation Continuous Wave Displacement Sensor Based on STM32H743
Han Yuan, Zheng Gang, Zhang Xiongxing, Sheng Qiming, and Bai Lang
Aiming at the problem of low signal sampling rate and signal processing speed of linear frequency modulation continuous wave displacement sensor, this paper designs a fiber optic displacement sensor with a Fabry?Perot interferometer structure based on STM32H743 chip. The optical fiber displacement sensor uses sawtooth wave to modulate the frequency of the laser, and selects STM32H743 as the core processor to improve the signal acquisition and processing speed. Its main frequency is 400 MHz, and the analog to digital converter maximum conversion rate is 4.5 MHz. Combined with frequency modulated continuous wave laser interferometry, the displacement of stainless steel pipe with fixed cavity length and moving target is measured. Experimental results show that in the 200?400 mm measurement experiment, the standard deviation of the measurement result of 1 mm/s is less than 3.3 nm, and the linear fitting coefficient in the range of 600 mm is above 0.99998, which has a good application prospect in the field of displacement measurement.
  • Jun. 09, 2021
  • Laser & Optoelectronics Progress
  • Vol.58 Issue, 11 1112007 (2021)
  • DOI:10.3788/LOP202158.1112007