Static Image Compression Algorithm For Wireless Image Sensor

Low-power wireless image sensor has a wide range of application prospects in the field of security and surveillance, urban management and heritage conservation. With the development of hardware, the quantity of image pixels that a camera can capture is getting bigger and bigger, therefore the transmission load of the sensor network becomes large, in order to prolong the lifetime of the sensor node which is powered by a battery, we need to compress the image to reduce the amount of transmitted data and the power consumption of a sensor node..The wireless image sensor takes photos of a monitored scene continuously every definite time interval, and transmits the captured images to the server via wireless network, the time interval is very long, it may be a couple of hours or even days. The transmission load in a wireless image is too large, the main work of this article is to study the interframe image compression algorithm in specific application scenarios for this problem. However, because of the long time interval, the light intensity and the lens focus may change greatly. Traditional MPEG motion picture compression algorithm which needs to be simplified is too complex for the image sensor. After analysis, the difference of the adjacent images is mainly reflected in the difference in brightness and zoom effect. Therefore, this paper will focus on these two points to do algorithm research and design, mainly include the following:Proposed an evaluation method for reference image selection based on brightness uniformity for images which have different quality. Since the exposure amount of the LED light each time is different, some captured images are overexposed in center area while the surrounded area is underexposed. This article proposed a liner intra block brightness adjustment method based on overlapping blocks to make the brightness of this kind of image uniform, to eliminate or reduce the boundary effects caused by blocks, we use weighted fitting between image blocks to achieve a continuous transition on the boundary region. Besides, we use histogram equalization to reduce the difference. The images will have centripetal movement because of the respiratory effects of the lens, this article proposed a motion estimation method based on centripetal movement to improve the effect of diamond search.By compressing the image data through some image processing such as brightness adjustment and motion estimation, we can reduce the amount of final transmission data, and the server can restore mages with a relatively good quality. The PSNR result is high in our demonstration system of wireless sensor network based on smartphones. The demonstration system has been applied in the Dunhuang Mogao preventive protection monitoring system.