Human Detection for Video Surveillance System

Detecting human beings more important in a visual surveillance system is crucial for diverseapplication areas including abnormal event detection,person identification, gender classification and fall detection for elderly people. Thefirst step of the detection process is to detect an object which is in motion. Object detection couldbe performed usingoptical flow, background subtraction, and spatio-temporal filteringtechniques. Once detected, a moving object could be classified as a human being usingtexture-based, shape based,or motion-based features.In this paper, we presents a real time Human detection algorithm based on HOG(Histograms of Oriented Gradients) features and SVM (Support Vector Machine) architecture. Motion detection is used to extract moving regions, which can be scanned by sliding windows;detecting moving region can subtract unnecessary sliding windows of static background regionsunder the surveillance conditions, then detection efficiency can be improved. Every slidingwindow is regarded as an individual image region and HOG features are calculated as classifiedeigenvectors. At last, the detected video objects can be categorized into predefined groups ofhumans and other objects by using SVM classifier. After Experiment, the results from videos areprovided with high accuracy of Human Detection.