An Algorithm for Detecting the HOG Features of Head and Shoulder of Football Players Based on SVM Classifier

Aiming at the problems of the UAV video, football players block each other seriously, the illumination changes obviously, and the image shooting quality is not high etc. This paper proposes an algorithm based on SVM classifier is proposed to detect the hog features of the head and shoulder of football players. This method take the HOG characteristics of football players' head and shoulder as the research object, intercept the positive samples including the moving head and shoulder and the negative samples excluding the moving head and shoulder, training the SVM classifier with the HOG feature of the samples, and The foreground area is scanned in multi-scale to detect the dynamic moving target. The experimental results show that the method can achieve real-time processing, target detection and recognition accuracy.