Research On Algorithms Of Multi-focused Image Fusion And Distributed Image Compression Technology For Wireless Sensor Network

Data fusion and distributed data processing technology in Wireless Sensor Network(WSN) have been two of the major topics of WSN in21st century, and they have very broadapplication prospects in military, monitoring, health care, smart home and so on.Multi-focused image fusion and distributed image compression are the hotspots of them.Because there are different targets with different distances in a scene, the image sensor can’tfocus on them at the same time. It’s hard to obtain the whole detailed picture of the specificscene. With Multi-focused Image Fusion technology we can get a complete and clear image ofthe scene in order to raise the recognition rate. In addition, the processing capability andenergy of a single sensor node is limited, centralized processing will soon run out of theenergy of the sensor node and reduce the lifetime of the network. With distributed imagecompression technology, we can greatly improve energy consumption balance, so as toprolong the lifetime of the whole network. This dissertation mainly research on multi-focusedimage fusion algorithm and distributed image compression algorithm. An improved fusionalgorithm based on directional region-characteristic and the distributed image compressionalgorithm based on SPIHT are brought up, and demonstration platforms are designed for themin this dissertation. The principal tasks and improvements are as follows:1. Expound the research background of data fusion technology and distributed dataprocessing technology in Wireless Sensor Network. Then mainly analyze the features and thetraditional algorithms of multi-focused image fusion and distributed image compression, andexpound the application and principle of wavelet transform.2. An improved fusion algorithm based on directional region-characteristic is brought upin this dissertation. This algorithm fully considers the principle of directional correlation inthe coefficients of wavelet transform and chooses the fused coefficients according to thedirectional region-characteristic. The characteristics that the difference of high-frequencysub-bands in multi-focused images is obvious and the wavelet transformed image hasdirectional correlation are fully used to improve the effect of multi-focused image fusion.3. A distributed image compression algorithm based on SPIHT is brought up in thisdissertation. This algorithm separates the channels and segments of the color image, andfulfils the coding task on different nodes in parallel. It overcomes the uneven distribution ofbig energy consumption in normal distributed coding algorithms. Compare with centralizedcompression algorithm, this algorithm prolongs the lifetime of Wireless Sensor Network.4. Demonstration platforms are designed for multi-focused color images fusion and distributed image compression. Algorithms brought up in this dissertation are used in theseplatforms and the effect is good. A general fusion processing platform is also designed basedon the multi-focused image fusion platform. This platform combines the functions andadvantages of multi-focused image fusion and image stitching, maximizes the fusion imageand reduces the transmission data, so it has great practicality.