ECLC: Edge character and latency connection enhanced inter-frame difference algorithm for real-time outdoor surveillance

This paper presents a technique, ECLC, for improving the accuracy and robustness of the inter-frame difference motion detection algorithm at a very low computation cost, which surpass the latest classical background subtraction algorithm ViBe. It incorporates several innovative mechanisms: edge characteristic info is introduced for more accurately identifying the foreground model with gradient info consideration; neighbourhood mean filter based latency connection is carried out to strengthen the foreground pixel's spatial correlation. Thoroughly experiments have been carried out, and efficiency figures showed that our algorithm can get exact moving object from complex background accurately like GMM and ViBe at a very low computation cost, which is even lower than ViBe. And also ECLC can solve some other typical problems such as ghost, shadow and illumination changes.