Sar Image Compression Based On Wavelet Transform And Recognition

The time-frequency properties of wavelet and fundamental theorem of image processing have been discussed in this thesis based on wavelet, fuzzy cluster analysis. Some research results made by myself are included, which are mainly through NSFC project 'Object recognition in satellitic and aerial image based on wavelet theorem' NSFC key project of 'Improving the radar imaging quality by signal processing methods' and the international collaboration project with ENST Paris France. We analyzed and researched the principles of SAR (Synthetic Aperture Radar) imaging, context features and the difference between the SAR image and the optic image used in many recent documents. The SAR image has very important use in many domains because SAR can work under any times and any weathers. It has been paid a great attention to in developed countries. In the research of SAR imaging, the problems such as SAR image transferring, restore and recognition become very challenging questions. Owing to the remarkable difference between SAR images and optic images, the SAR image processing is also concerned with the electromagnetic theorem and the modem mathematics, forming a front research area in the domain of information engineering0 The SAR image not only has the features of optic image, such as geometric feature, algebra feature etc, but also has some electromagnetic features, so there are lots of information included in the context feature of it. This feature arouses new problems to the SAR image compression and recognition. How to keep this context feature when the compression rate is improved? How to obtain more useful information using the electromagnetic features in SAR image? The main purpose of this thesis is to try to answer these questions. It is possible to reconstruct the 3D digital map from a single SAR image by using electromagnetic features in SAR image. Reports from USA and France said that the work provides very rich resources of 3D digital map for the departments of topography and military, also can increase the information resource for the SARv image compression and recognition. One of the problems is how to keep the context information in SAR image compression, transferring and recognition. The research on SAR image compression and automatic recognition is just underway in the world, but some good results have been presented, such as SPAM for the recognition of aerial image, SIGMA for the recognition of urban aerial image, MESSIE for satellitic remote image, and ICARE for the recognition of SAR image etc. The work by H.Maitre, J.M.Nicola, I.Bloch, F.Tupin etc, in the Image processing laboratory of Ecole Nationale Supereuse des Telecommunications de Paris are more attractive. The main works of SAR image processing include geometric modifying, image gray level modifying, image enhancement, denoise (speckle and band noise) and image compression. The main theoretical researches are recognition, segmentation, 3D-reconstruction etc. The main contributions of this thesis is as follows: 1. Improved four-tree image compression algorithm based on wavelet transform theorem and used lifting system wavelet algorithm to the SAR image compression. The later method makes the wavelet algorithm more rapid. The lifting system wavelet algorithm (t...