Research on Target Recognition Algorithm Based on OpenCV

In this study, in the HSV color space converted from RGB color space, the target recognition was carried out through different color channels and OpenCV. First, in this study, a color recognition and quantity estimation model was established to find the possible location of the target within the HSV range of different colors, reduce the noise and irregular edges in the image, and draw a histogram to visualize the distribution of the number of targets in all images. Secondly, this study establishes a target position recognition model, which recognizes different colors by setting a specific HSV threshold, then uses cv2.findContours to find the target contours, calculates its centroid as the target center point for each contour, and then converts the coordinates of the center point to draw a two-dimensional scatter plot and a scatter plot heat map, which visually shows the geometric position and distribution density of the target in the image. Finally, a target state evaluation model was established, which extracted the red and cyan parts of the target respectively, and calculated the ratio of the red pixels in each target to the sum of the whole target pixels, so as to plot histograms, box plots and cumulative distribution plots to show the distribution of states. In summary, this study uses OpenCV to achieve target color recognition and quantity, position and state estimation in the HSV color space, which provides a reference for other studies.