Abstract: Due to the CMOS camera used in Yunnan Observatory 53cm binocular laser ranging system, space target ranging echo rate and ranging success rate has been greatly improved. On the basic of analyzing the CMOS digital image information of laser ranging, an optimized image processing and laser light point feature extraction algorithm is proposed, which has efficiently achieved laser light point recognition and target centroid coordinate extraction. Median filter and average filter remove noise in the image; the Otsu method calculates the thresholds to realize image binarization and the Roberts operator is selected to detect the edge; linear fitting using least squares method solves the light point coordinates and calculates the position deviation. The results have important implications for further research in the application of CMOS camera in the 53cm binocular laser ranging system.