Motion detection is a very import process and the quality of detection result is critical to the further processing.This paper researches on existing motion detection algorithms based on background subtraction. And this paper also presents a simple model for performance evaluation of motion detection algorithm based on background subtraction, with the use of principle of the traditional performance evaluation methods of motion detection algorithm. This model bases on the following truth: The ground truth can be gotten by manual or by other reliable procedure. This paper measures several motion detection algorithms based on background subtraction and the experiment results show that this model can quantitatively show that which algorithm do better on special detection environment. What's more, this paper implements the enhanced W4+ algorithm according to W4+ algorithm. Enhanced W4+ inherits the advantages of W4+ and we add performance evaluation model of motion detection algorithm to it. By showing people quantitative results, we can conclude that enhanced W4+ do better job than W4+. Finally, this paper setups input image series and their corresponding reference library (Ground Truth). After that, this paper does some experiments on W4+ and enhanced W4+. Experimental results show that which algorithm has better detection results quantitatively.