Single-object-based segmentation and coding technique for video surveillance system

摘要

We present an adaptive object segmentation based onnscene change detection techniques. First, the algorithm adaptivelynadjusts to skip the number of frames to cope with the amount ofnmotion displacement to cover the entire object shape. Next, the spa-ntial processing consists of noise removal and boundary smoothingntechniques to discard background content as well as to remove thenbackground noise, to obtain a segmented object. To evaluate thenquality of segmentation, both of the standard benchmarks and cam-nera imaging are employed. Results show that the proposed algo-nrithm can achieve low error ratios under various standard sequencentesting. The segmentation algorithm combines with MPEG-4 codingnfor camera imaging of video surveillance. This system can success-nfully demonstrate real-time video surveillance, as the frame rate isn10 to 30 using CPU-based software implementation.