Aiming at the requirement of laser dispersed quenching (LDQ) quality inspection for metal surface, a visual inspection system based on OpenCV was designed to measure the quenched spot size and detect the surface texture. In-depth study of the relevant image processing algorithms, Canny edge detection algorithm has been improved to achieve the selective extraction of image contours, the geometric dimensions of the laser quenching connected domain are obtained, and the gray level co-occurrence matrix is used to calculate the texture information of quenched spot. According to the quenched spots produced by LDQ under three different energy input backgrounds, a dual threshold determination method was proposed to realize the quality inspection. The average detection time was 10.1ms, and detection accuracy was above 92.5%.