Efficient Implementation of GMM for Image Segmentation

[摘要] Identification of Background is very important feature in Image Processing. Detecting the Background from its Foreground is unique feature used in many applications such as Military, Traffic monitoring and Video Surveillance etc. This separation can be achieved by background subtraction methods but they are not that efficient. The most commonly used approaches are Statistical model and GMM. Among these statistical modal is not applicable in multimodal background in which objects shows repetitive motion. The main aim of this paper is to separate background and foreground using Gaussian Mixture Model(GMM), background identifying algorithm and designing soc Using Mat lab and Cadence tool which gives high accuracy and less power consumption.

[关键词] GMM;Statistical Method;Background Detection;Image Processing