In the field of intelligent video surveillance, in order to realize the real-time detection of moving pedestrians in surveillance video, this paper introduces a method to realize pedestrian detection by using the gradient histogram (HOG) feature of image and support vector machine (SVM). In this method, the local gradient histogram of positive and negative sample images is extracted, and the support vector machine is used for sample training to obtain the pedestrian classifier. Finally, the trained classifier is used for pedestrian detection. Experimental results show that this method can effectively detect moving pedestrians in surveillance video, and has good detection effect.