The Low-Complexity Video Coding Algorithm For Wireless Video Sensor Networks And Implementation Based On ARM Platform

Since 21th, with the rapid development of computer technology, information processing technology, semiconductor technology and networks technology, the human world has entered the time of information. As the same time, the Wireless Video Sensor Networks (WVSN) are attracting a considerable interest. WVSN have many advantages and broad applied future. It also has potential using value in many fields, such as military affairs, industrial construction, city management and video surveillance system and so on.WVSN have some remarkable characters, for example, the power of network nodes is limited; the bandwidth for communication is narrow; high requirement for the processing speed. So, the tradition video coding standard cannot be used in WVSN. MPEG-4, H.263, H.264 video coding standard, are all achieved based on motion estimation and motion compensation which have a great deal of computation. it is not practical to perform such algorithm on the resource constrained nodes.In this paper, based on the specific requirements for Wireless Video Sensor Networks (WVSN), a motion detection based video coding algorithm with low computational complexity has been proposed. It only encodes the motion objects in current frame and outputs the object-oriented bitstream. Experimental results show that the speed of the proposed algorithm is about three times higher than H.264 all I-frame encoder while the coding performance can be improved about 2dB. Compared with H.264 baseline encoder, though the coding performance of the proposed algorithm is a little worse than that of H.264 baseline encoder, the encoding speed has increased at least 8 times higher on average. In conclusion, the proposed algorithm has achieved good tradeoff between the coding performance and encoding speed, which can meet the requirement of WVSN to some degree.In this paper, analyzing the structure of algorithm and considering the characters of the embedded system, several methods for optimization are presented in some aspects, including algorithm, memory, advanced language and assemble language.Finally, the present algorithm is realized on the ARM embedded platform. The results show that, the speed is improved remarkably, which can meet the requirements of the real time processing surveillance video sequence with CIF format.