Moving human detection algorithm based on Gaussian mixture model

In order to realize accurate detection of moving human in complex and cluttered scene and eliminate the influences of moving shadow, an adaptive moving human detection algorithm based on Gaussian mixture model is presented. This algorithm employs Gaussian mixture distribution in modeling background of each pixel and an on-line approximation to update the model. The moving human region is detected by background subtraction. The shadow detection and elimination is implemented in the HSV space, and finally morphological operations are introduced to eliminate noise and reconstruct moving human regions. The results show that the method models stable background, eliminates shadow, and has a good detection result.