Real-time periodic motion detection and background estimation

Motion detection is a process to detect a changing in position of an object based on its surroundings. This paper, therefore, will present a periodic motion detection and background estimation. This method focussed on outdoor illumination condition. The method used is sturdy to illumination change effect, change in background and noise. Background subtraction is used in this method and the background image is estimated every 0.8 second using timer if the sum of absolute different (SAD) is less than the motion threshold. The input image is uminance normalize before the background subtraction. The results were converted into a binary image by autothreshold and the results were enhanced with dilation and erosion. The blobs were created for each motion object. Experimental results of using a background image estimated by periodic background estimation demonstrated their sturdiness and effectiveness in real-time background subtraction.