Laser-Visible Face Image Translation and Recognition Based on CycleGAN and Spectral Normalization.

The range-gated laser imaging instrument can capture face images in a dark environment, which provides a new idea for long-distance face recognition at night. However, the laser image has low contrast, low SNR and no color information, which affects observation and recognition. Therefore, it becomes important to convert laser images into visible images and then identify them. For image translation, we propose a laser-visible face image translation model combined with spectral normalization (SN-CycleGAN). We add spectral normalization layers to the discriminator to solve the problem of low image translation quality caused by the difficulty of training the generative adversarial network. The content reconstruction loss function based on the Y channel is added to reduce the error mapping. The face generated by the improved model on the self-built laser-visible face image dataset has better visual quality, which reduces the error mapping and basically retains the structural features of the target compared with other models. The FID value of evaluation index is 36.845, which is 16.902, 13.781, 10.056, 57.722, 62.598 and 0.761 lower than the CycleGAN, Pix2Pix, UNIT, UGATIT, StarGAN and DCLGAN models, respectively. For the face recognition of translated images, we propose a laser-visible face recognition model based on feature retention. The shallow feature maps with identity information are directly connected to the decoder to solve the problem of identity information loss in network transmission. The domain loss function based on triplet loss is added to constrain the style between domains. We use pre-trained FaceNet to recognize generated visible face images and obtain the recognition accuracy of Rank-1. The recognition accuracy of the images generated by the improved model reaches 76.9%, which is greatly improved compared with the above models and 19.2% higher than that of laser face recognition.

Electro-optic modulation aberration correction algorithm based on phase difference compensation.

In order to achieve wide field-of-view, high-resolution LIDAR, a gating imaging structure combining an electro-optic crystal and an electron multiplication CCD is constructed. According to the index ellipsoid theory, a 3D ray tracing model is established to explore the principle of electro-optic modulation. The field-of-view and interference intensity distribution of the LiNbO3(LN) crystal electro-optic modulation are studied by using the proposed model. In order to solve the problem that the interference light intensity at the edge of the field-of-view of crystal electro-optic modulation is not homogeneous, we study and propose an electro-optic modulation aberration correction algorithm based on phase difference compensation. The corrected interference light intensity at the edge of the field-of-view increased from 0.87 to 0.94. Finally, a LIDAR imaging simulation system is established to image the target at a distance of 2000 m. The results show that under the condition of a 1° wide field-of-view, the imaging accuracy of the system is 5 mm, and the average imaging error introduced by the crystal electro-optic modulation is less than 2 cm.

Photoelectric Target Detection Algorithm Based on NVIDIA Jeston Nano.

This paper proposes a photoelectric target detection algorithm for NVIDIA Jeston Nano embedded devices, exploiting the characteristics of active and passive differential images of lasers after denoising. An adaptive threshold segmentation method was developed based on the statistical characteristics of photoelectric target echo light intensity, which effectively improves detection of the target area. The proposed method's effectiveness is compared and analyzed against a typical lightweight network that was knowledge-distilled by ResNet18 on target region detection tasks. Furthermore, TensorRT technology was applied to accelerate inference and deploy on hardware platforms the lightweight network Shuffv2_x0_5. The experimental results demonstrate that the developed method's accuracy rate reaches 97.15%, the false alarm rate is 4.87%, and the detection rate can reach 29 frames per second for an image resolution of 640 × 480 pixels.

Simulation and Experimental Research on a Beam Homogenization System of a Semiconductor Laser.

Aiming at the application of laser active imaging detection technology, this paper studied the beam homogenization system of a semiconductor laser based on a homogenizing pipe. Firstly, the principle of the homogenizing pipe was introduced. Secondly, the homogenization effect, which was influenced by several geometric parameters (aperture size, length, and taper) of the homogenizing pipe using the optical design software, was simulated for the fiber-coupled semiconductor laser. Finally, according to the simulated results, a laser illumination system composed of a fiber-coupled semiconductor laser, a homogenizing pipe, and an aspheric lens was designed, which can obtain a rectangular uniform light spot in a long distance. The effectiveness of the illumination system was verified by simulation and experiment, respectively. Simulation results suggested that the uniformity of the spot at a distance of 20 m was 85.6%, while divergence angle was 10 mrad. The uniformity of the spot at a distance of 120 m was 91.5%, while divergence angle was 10 mrad. Experimental results showed that the uniformity of the spot at a distance of 20 m was 87.7%, while divergence angle was 13 mrad. The uniformity of the spot at a distance of 120 m was 93.3%, while divergence angle was 15 mrad. The laser illumination system designed in this paper was simple and easy to assemble, and has strong practicability. The results in this paper have certain reference value and guiding significance for the homogenization design of semiconductor lasers.

The Effect of Light Source Line Width on the Spectrum Resolution of Dual-Frequency Coherent Detection Signals.

In this paper, the power spectrum resolution problem of dual-frequency coherent mixing signals is analyzed when the Doppler frequency difference is small. The power spectrum function formula of the four optical coherent mixing signals is obtained using statistical theory and the Wiener-Khinchin theorem. The influence of delay time and light source line width on the power spectrum of dual-frequency coherent signals is analyzed using this formula. The results show that delay time only affects the peak of the power spectrum of the coherent signal. An increase in the line width of the light source broadens the signal power spectrum and reduces the peak value. The necessary condition for distinguishing the Doppler frequency difference is that the theoretical Doppler frequency difference is greater than 1/5 times the line width of the light source.