Research On Coding Of Remote Sensing Images Based On Integer Wavelet Transform

Recent advances in the field of remote sensing require us to compress abundance of remote sensing image data. Wavelets have been successfully used in compression in all kinds of image. The remote sensing image has high resolution, great information quantity and high rate, so the compression method not only need high compression ratio, low distortion, but also need fast compression speed and high reliability. The compression method based on wavelet transform to compress the image data is adopted, after compared the general compression methods and coding criterions. The research on compression of remote sensing images based on integer wavelet transform are explored in this paper.The image real time compression is an important question in remote sensing technique, but the general compression methods haven't achieved the best compression result, therefore developing a real time compression method which is suitable for remote sensing image are of significant practical and commercial interest. Although the techniques of data storage are developing and the bandwidth of channel is wider than past, the demand for compression of image data is rapidly growing so that to save the space of storage and improve the use ratio of channel. Comparing with lossy image compression, the lossless image compression is more difficult and its development is slower. Now, image compression method based on wavelet is the branch which development is fast in image compressing field. Lifting scheme can design the wavelet transforms that map integer to integer, which provides the effective tool for studying the image compression coding. On the other hand, the method for wavelet coefficients coding is a key technique to implement image compression, which not only affects the effect of compression, but also the quality of reconstructed image and the time of coding and decoding.First, the theory of wavelet transform was studied, and it was found that the traditional wavelet transform uses convolution, and it makes the transform process complicated, and the results are floating point numbers. Therefore lifting scheme is used in image compression to decompose image. When wavelet transform is applied to image compression, the chosen wavelet base affects the speed of transform and the quality of the reconstructed image. It is very important to research the correlation between the wavelet base characters and image compression. This thesis analyzes the correlation through extensive experiments, and presents the principles of wavelet base choice in image compression. Based on the classical wavelet coding methods, a new coding algorithm of fast image compression is presented. The improved SPIHT algorithm drastically reduces both the memory requirement and the time consumption.The method based on scalable quantized significane testing schemes is an important type of coding for wavelet coefficients, which characters are low complication, good performance, and so on. Algorithms, such as the Embedded Zerotree Wavelet algorithm(EZW), Set Partitioning In Hierarchical Trees(SPIHT), which are classical algorithms based on wavelet for image compression, are belong to this type of method.Wavelet Analysis is a novel theory developed during past 20 years. The technique is the great achievement in Harmony Analysis field for about half of this century and is viewed as a milestone in the development history of Fourier analysis. As a new theory of signal analysis, Wavelet Transform appears the good localization property in time (space) domain and frequency domain, so we obtain an effective mathematicl tool and use it to lots of research fields such as image processing, pattern recognition etc. Based on the theory of Wavelet Transform, this thesis focuses on researching the compression coding algorithm for the remote sensing image.The research works of the thesis are listed as following:(1) Summarized the principles and development history of image compression, and briefly compared the traditional image coding technique with the 2nd generation image coding technique.(2) Surveyed Wavelet Transform from the view of multi-resolution analysis, stated the relevant basic concepts such as wavelet function, discrete wavelet transform, continuous wavelet transform and fast speed algorithm of wavelet transform, and pointed out the advantages of Wavelet Transform by comparing with traditional Fourier Transform.(3) Based on HAAR Wavelet and S Transform, introduced the Integer Wavelet Transform with the characteristic of the lifting scheme, which establishes the theoretical basis for the following research on compression coding for the remote sensing image.(4) Analysed the SPIHT algorithm comparing with EZW algorithm and simulated it in test to select wavelet basis. Further the thesis demonstrate that the wavelet basis will affect on the compression performance for same application of remote sensing image. (5) Discussed SPIHT algorithm in detail. To confront the shortcoming of SPIHT algorithm when it was applied to compression of remote sensing imge, we proposed an improved algorithm named as SPIHT- INT-RSI which encoded the residual data better. The simulation results showed the SPIHT-INT-RSI algorithm was effective.(6) As one kind of remote sensing image, the multispectral image has attracted more and more attention. By using the existed achievements in compressing multispectral image area, this thesis utilized 3D-SPIHT algorithm and its improved algorithm into compression coding for these images and the simulation results showed the good effect.