In the comparison and selection of high-sensitivity detectors, the sensitivity, imaging contrast and resolution are important indicators to evaluate the imaging capability. The standard test method of resolution has a high demand for the environment and complicated and inconvenient. In this paper, we introduce a low-light-level resolution testing and comparing system to facilitate and compare and evaluate the sensitivity, imaging contrast and resolution of different high-sensitivity detectors, and provide the basis for the selection of the detector. The pattern of the target plate is designed. The target plate is made of an optically chrome-plated plate and the inner wall of the enclosed system where the optical system is located is coated with a black diffuse paint to reduce the interference of the stray light to the test. According to the sampling theorem, in order to ensure that the narrowest stripes after imaging meet the resolution testing requirements of the detector which has the smallest pixel, the narrowest stripe width of the imaging on the photosensitive surface should be less than 1/2 of the smallest pixel. The use of focusing makes it can still be clearly imaged when there is a small assembly error, but the focusing should not affect the test results. Finally, the system is used in an experiment, testing and comparing the resolution and sensitivity of three detectors, and verify the effectiveness and convenience of the system.
The centroid detecting accuracy of the stripe has an important influence on reconstruction and configuration of object structure in structured light 3D scanning measurement system. This paper analyzes the influence of several factors on the accuracy of stripe centroid detection. The pictures are grouped according to the peak signal-to-noise ratio and the stripe width. The gray-scale matrix is read out by Matlab, and then the centroid position is calculated by using the gradient centroid algorithm for the gray scale image by subtracting the threshold from the window. Finally, the accuracy of each group of images is evaluated by its standard deviation. The standard deviation of the group image is taken as the evaluation criterion, and the influence of PSNR and the stripe width on the accuracy of centroid detection is obtained. The experimental and simulation results show that the narrower stripes are more able to achieve higher accuracy; the accuracy is improved as the signal to noise ratio increases and multi-frame stacking can significantly improve the detection accuracy. The variation curves of the factors on the accuracy are given in the text. At the same time, the phenomenon that the measured value is fluctuating under a peak signal-to-noise ratio is analyzed by simulation. The results show that the fluctuation is consistent with the normal distribution.