Research On Moving Object Detection Algorithm In The Embedded Platform

80% of all the information that people grasped was obtained from the image, the image-related information plays an extremely role in human life, what`s more, it can not be replaced by other information. Before people get information about the image, the image should be processed firstly, thus came with the digital image processing technology. Currently, digital image technology has become the object of computer science, statistics, engineering, physical sciences, chemistry and even social studies.Moving object detection is an image segmentation method that detected the variation movement area from the image sequence and extract the moving objects’ information which people interested from a video sequence. Moving target detection technology use the image processing method for moving object parsing in a sequence of images, achieve the goal of quantitative measurement and qualitative analysis. Moving target detection the basis of moving target tracking, analyzing and understanding, for it will directly affect the accuracy of the test results and the subsequent image stability operations.However, due to the interference brightness changes, shadows moving targets, noise and other factors, the existing moving object detection methods are still computationally intensive, time-consuming, high energy consumption. Therefore, how to detect moving targets quickly and accurately is an important research topic in object detection technology.In addition, the emergence and development of embedded systems technology plays a vital role in building the information society, embedded systems has advantages of good reliability, high performance and portability. Embedded systems are oriented products, applications and users, therefore it will have more advantages and practical significance when it bind with specific applications. Currently, embedded systems are widely used in industrial control, traffic management, information appliances, POS network, environmental engineering, defense and aerospace and other fields. With the improvement of the performance of embedded processors, combining with the image processing technology has become a trend, and in many areas of the smart home, smart city, intelligent transportation has been widely used.In this paper, we proposed a new algorithm which based on Vibe and shadow removal algorithms, to solve the problem of the existing traditional modeling methods in a complex environment can not extract the moving target accurately and completely. Firstly, we described the vibe algorithm, and make a improvement and optimization on it, then extract the target prospect which contain shadows by using the improved vibe algorithm.Secondly, extracted the background by using K-means clustering algorithm combined with the vibe algorithm. Lastly, removed the shadows in foreground objects by using the HSV color space shadow removal algorithms, then we got the accurate foreground object.Finally, we made the experimental vertification of the proposed algorithm on the embedded platform. The experiment results show that the proposed algorithm can adapt to the dynamic backround with noise and shadows, and the moving objects of detection is integral and clear.