Improved Gaussian mixture model based moving target detection

To solve the problems that traditional Gaussian mixture model can not match pixels well and updates parameters slowly,an improved Gaussian mixture model method is proposed.The presented algorithm makes more accurate matching methods by calculating Mahalanobis distance in terms of model matching operation judgments.What's more,parameter updating takes advantage of the concept of the number of matches and the match,which can update Gaussian mixture model parameters in real time and avoid complex calculation,so that the new model can establish background model more quickly and make it more adaptive to the complex background.The experimental results demonstrate that the improved Gaussian mixture model proposed is more instantaneous and accurate when detecting moving target.