The Research Of Pixel-level Image Fusion

Image fusion means to synthesize multiple source pictures and get more accurate, more comprehensive and more reliable information about same scene or object. Fused image should be convenient for the following image analysis, recognizing, target detecting and tracing. It takes full advantage of redundant or complementary information from source pictures, and not be identical to common image enhancement. It is a new technology in the area of computer vision and image comprehension. Image fusion technology is generated from military area, such as missile guide, autonomous robot, unmanned scout flight and so on. At the same time image fusion technology is applied in a wide variety of fields such as medical diagnosis, remote sensing and industry manufacture. In this dissertation, we focus on the method of image fusion and evaluation of its quality. The concept, purpose, application of image fusion are also involved. The routine fusion methods are introduced in detail, such as simple fusion method ( pixel average, maximal or minimal pixel selection), the multiresolution image fusion techniques based on multiscale pyramid decomposition, and the image fusion method based on wavelet transform,and develop a method to determine coefficients in pixel average algorithm with principle component analysis(PCA).Likewise, evaluation for fusion image is vital to fusion system. Some qualitative and quantitative criterions are also introduced in detail. At last, after studying of image formation model and modifying the model, a novel image fusion algorithm is proposed ――we can educe the fused image from the model based on non-multi-scale decomposition, with the help fuzzy math and EM algorithm. A series of experiments on image fusion are given in this dissertation. The experimental and calculation results show that the fusion algorithm we propose is effectual.