遥感图像基于小波变换压缩系统的快速实现

【作者】 王沛；

【导师】 黄廉卿；

【作者基本信息】 中国科学院长春光学精密机械与物理研究所 ， 光学工程， 2000， 博士

【摘要】 图像压缩是当今数字图像处理领域关键技术之一，在很多方面均有广泛的应用。过去二十年以来科学研究的发展证明了这种技术的重要意义。早期图像压缩的研究多集中在离散余弦变换上，近期压缩算法则更多集中在多分辨率技术上，尤其是小波变换。 遥感图像编码压缩是遥感技术的一个重要组成部分。本文针对遥感图像分辨率高、相关性弱、冗余度小、纹理多的特点，提出了一种基于小波变换遥感图像压缩编码系统快速实现方案。 本文首先从数学分析的角度研究了小波变换的理论基础、特点和性质，介绍了小波变换在应用中的具体步骤及其流程图。在对遥感图像进行小波变换后，根据人类的视觉、生理及心理特点，对低频子图象和高频子图象分别进行不同策略的量化和编码处理。其中，低频子图象用插值差分脉冲预测编码调制方法进行编码，而高频子图象用自适应量化和算术编码方法进行编码。为了加快程序运行速度，用汇编语言设计了图像压缩系统软件的关键代码，经过MMX指令优化后，将之放置于操作系统的Ring 0模式下，有效提高了图像压缩速度及压缩系统的稳定性。 本文研究的压缩系统在压缩比4：1和8：1时的压缩质量完全可以满足遥感图像的压缩要求，而所用的压缩时间比起JPEG2000和JPEG压缩所需的时间提高了一个数量级。 小波变换及其应用是当今国内外研究的热点。本文研究方法将对应用小波变换的各类图像数据压缩处理领域具有一定的理论指导和实际应用价值。更多还原

【Abstract】 Image compression is one of the key techniques in today’s digital image-processing field. It has wide applications. The considerable body of research over the last two decades has shown the significance of this technique. The early research on image compression mainlysfocused on the discrete cosine transform(DCT). While the recent research mostly focus on multi-resolution techniques, particularly the wavelet transform.Image compression plays a significant role in the field of remote sensing technology. Considering many important characteristics of remote sensing images, such as high resolution, little correlation, low redundancy, rich texture, etc., we have proposed a WT-based, fast encoding image compression system for remote sensing images.At first, the theory basis, characters and property of wavelet transform are analyzed from the view of mathematical transformation , the practical steps and their flow charts are provided in this paper. After a remote sensing image ,has been decomposed by wavelet transform, low frequency sub-image and high frequency sub-images can be quantified and encoded with different techniques that match the human visual , physiological and psychological characteristics. In essence, the coding algorithm involves performing RIDPCM coding method on the wavelet coefficients of low frequency sub-image and performing adaptive quantization and arithmetic entropy coding method on the wavelet coefficients of high frequency sub-image. In order to accelerate procedure’s operation speed, the key codes of the software in image compression system are designed in assembler language. Then the procedure is optimized by MMX instruction and put at the mode ofRing 0 level. Therefore, the image compression speed and the stability of the compression system are enhanced.With this compression system, the compression quality can meet the compression demands of remote sensing images when the compression ratio reaches 4:1 or 8:1, but the compression time this system has been increased by an order of magnitude, compared with that of JPEG 2000 or JPEG.Wavelet transform and its application are hot points in recent domestic and international studies. The study method in this paper has some theory direction and practical application value for various fields of image data compression processing using wavelet transform.更多还原