INTERNATIONAL JOURNA L OF ENGINEERING SCI ENCES & RESEARCH TECHNOLOGY Video Frame Extraction Process on Canny Edge Detector Using Gmm Algorithm
T. Arumuga Maria Devi, N. Rekha · 2013
This paper presents a technique for motion detectio n that incorporates several innovative mechanisms. For example, our proposed technique stores, for each pi xel, a set of values taken in th e past at the same location or in the neighborhood. It then compares this set to the curr ent pixel value in order to determine whether that pixel belongs to the background, and adapts the model by choosing randomly which values to substitute from the backg This approach differs from those based upon the cla ssical belief that the oldest values should be repl aced first. Finally, when the pixel is found to be part of the background, its value is propagated into the backgr ound model of a pixel. We describe our method in full details (incl uding pseudo -code and the parameter values used) and compare it to other background subtraction tech niques. Efficiency figures show that our method out performs art methods i n terms of both computation speed and detection rat e. In this project, we propose a universal sample -based background subtraction algorithm, called moti on detection, which combines three innovative techniqu es. First, we propose a classification model seco explained how motion detection can be initialized w ith a single frame. Finally, we presented our last innovation: an