Abnormity recognition and extraction method based on carbon fiber connector image processing

The invention relates to the technical field of image processing, in particular to an anomaly recognition and extraction method based on carbon fiber connector image processing, and the method comprises the steps: obtaining a grayscale image of the surface of a connector; acquiring color information of each pixel point in the grayscale image, and acquiring color consistency of the corresponding pixel point based on the color information; according to the color consistency and the gray value of each pixel point, obtaining a fluctuation mean value in different preset directions; screening a target point based on the fluctuation mean value in each preset direction, and obtaining a direction threshold value in the corresponding preset direction according to the color consistency and the gray value of the target point; and according to the direction threshold value in each preset direction, obtaining a judgment value of the pixel point in the corresponding preset direction, and obtaining an abnormal defect area in