Surveillance of Video to Detect Moving Objects Using Background Subtraction Algorithm

Video Surveillance is needed for governments and government organizations to maintain social control, recognize and notice threats, and prevent criminal activity. In our approach, frames are taken from video and are converted to images. For each pixel in the image, the values of that pixel that are stored previously at the same location is taken and it is compared with the current pixel value to determine the background pixel. Then background pixel is removed and only foreground pixels that comprises moving object are detected. When the pixel is found as a background pixel, then its value is transmitted into neighboring pixel’s background model. By this way, current frame is compared with available frames that are already stored in database. Our approach have efficiency in detection rate and computation speed. The algorithm called Vibe is used to find foreground pixels. Then we apply Temporal Difference technique to find foreground pixels. Performance is evaluated and compared for these two algorithms. It also requires a minimum of one comparison and per pixel one byte of memory is needed