【作者】 王仁龙；

【导师】 郝燕玲；

【作者基本信息】 哈尔滨工程大学 ， 导航、制导与控制， 2009， 博士

【摘要】 由于雷达图像与视频信号不同,它要求15秒采集一幅图像,而不是像视频要求每秒25帧左右,所以雷达图像处理可视为静态图像处理。但是雷达图像作为航行数据记录仪(VDR)的重要组成部分,对图像质量的要求较高,不是主观看看就可以的,还有专门的客观评判标准；所以为了满足既有好的压缩比,又要保持高质量,压缩算法很关键。雷达图像分辨率高、信息量大,压缩技术不仅要求大压缩比和低失真度,特别要求可靠性高。在综合比较了国内外各种压缩算法、编码标准的基础上,针对目前几种小波算法,在编码过程中存在占用内存大、硬件实现难的问题,本文提出了基于块位长的小波编码算法作为雷达图像的压缩方法。首先基于人眼视觉特性,分析了目前的静态彩色图像的分量组合格式。通过了解三基色原理和静态彩色图像的分量组合格式,分析了常规图像需要的分量变换,针对雷达图像提出了不进行RGB到YCbCr的格式转换,不进行电平移位。接着研究了小波变换的理论基础,并结合信号处理知识对小波分析理论进行了系统的介绍和分析。文中先从Fourier变换引入小波,介绍小波变换和多分辨率分析,在多分辨分析框架的基础上,结合实际的数字信号处理,推导出Mallat算法。最后简述提升算法的基本原理,并给出了传统小波的提升实现；针对硬件实现选用LS(9,7)小波,并对其进行了改进。然后研究目前几种典型的基于小波变换的图像编码方法,分别是EZW、SPIHT和SPECK算法。通过对这三种编码算法的编码过程、原理的深入研究,探寻基于小波变换的图像编码的基本原理,最后针对SPECK算法的几点不足之处,结合SPIHT算法对SPECK算法进行了改进；改进算法给出了S集合块和I集合块的具体实现方式和分裂方式,这为编码实现提供了基础；在不继续增加熵编码的情况下,改进算法具有和SPECK原算法一致的编码性能,但是去掉了复杂的嵌套函数,编码方式简单明了；量化细化时,去掉了LIS链表再排序过程,加快了编码速度。针对当前的几种小波编码算法存在的不足之处,提出了两种基于块位长的小波编码算法。一种是基于块位长的无链表嵌入零块小波编码算法,另一种是基于块位长的无链表小波分块编码算法。这俩种算法都具有很高的信噪比。针对具体的硬件实现,考虑在编码过程中的内存占有量和所占用的时间,基于块位长的无链表小波分块编码算法具有更强的优势,所以在雷达压缩系统中采用基于块位长的无链表小波分块编码算法。最后给出了雷达图像压缩系统的硬件实现平台,详细说明了其工作原理,介绍了主芯片,说明了小波编码算法在硬件平台上的实现；接着给出了上位机测试程序,包括具体程序流程、上位机测试程序的主界面和配置参数界面；最后根据中国船级社颁发的《船载航行数据记录仪检验指南》中有关雷达图像压缩技术的国际测试标准,对雷达图像压缩系统进行测试,测试结果表明该系统完全达到了国际标准,并且长时间运行稳定可靠。更多还原

【Abstract】 A whole piece of radar image usually is required to be sampled every fifteen seconds, which is different from video image that generally has twenty-five frames per second, so it means radar images can be treated as static images. Radar image plays an very important role in voyage data recorder(VDR), and the quality of this kind of image needs to be relatively high which cannot be valuated arbitrarily by people’s eyes but should be estimated through some objective standard. So first and foremost, the desired compression algorithm should achieve an ideal compression ratio and keep a high quality of radar image.With the consideration that radar image with high resolution generally has huge amount of information, a large compression ratio, low distortion and especially high reliability are expected to be derived through the compression technique adopted. After an overall comparison of various compression algorithms and coding standards, a sort of wavelet coding algorithm based on block bit-length which can be used to compress radar image effectively was presented in this dissertation, which also aims at solving the problems such as occupancy of comparatively large amount of memory in coding process and difficulty in hardware realization which might exist in some kinds of wavelet algorithms used presently.According to visual characteristics of human eyes, of present static colorful image was analyzed. Along with the principle of the tricolor and the knowledge about the scalar component combination format of static colorful image, the scalar component transform for regular images was analyzed, also an innovative format transform without conversion from RGB to YCbCr was derived for radar image particularly and there is no voltage shift as well.Some research work focused on the fundamental of Wavelet Transform, and systematic introduction and analysis about wavelet theory linking with principles of signal processing technology was introduced additionally. In the dissertation, the concept of wavelet followed the introduction of Fourier Transform, Wavelet Transform and Multi-resolution Analysis were put forward subsequently. Then, Mallat algorithm on the base of Multi-resolution Analysis structure was acquired with some reference to digital signal processing theory. After that, basic principle for the improvement of the algorithm was discussed briefly accompanied with the realization of improvement method for traditional Wavelet Transform. Moreover, wavelet LS(9,7) was employed and some improvement was also implemented for the purpose about hardware realization.Following the above, a few of typical image coding arithmetics, EZW, SPIHT and SPECK, which based on Wavelet Transform were studied. The fundamental principle of image coding algorithm founded on the base Wavelet Transform was explored through the deep sights on the principles and coding processes of those three kinds of arithmetics, and some shortcomings in SPECK was overcome with the characteristics of SPIHT. The improved algorithm provided the concrete realization method and split method corresponding to the sets of S-block and I-block, which acted as a foundation for coding realization. The performance of this algorithm is identical with that of SPECK without increasing entropy codes and its coding process was simplified after getting rid of complex nest function. When the resolution was increased, the coding speed could be boosted with the elimination of the procedure where LIS link tables were resorted orderly.Tow sorts of wavelet coding algorithms connected with the concept of block bit-length were presented to overcome the drawbacks exists in some current wavelet coding methods. One is Zero Block Wavelet Coding algorithm without nested link based on block bit-length, and the other is called Wavelet Block Coding algorithm without link table based on block bit-length, both of them can provide comparatively high SNR, respectively. Under the consideration that both memory and time consumption in the coding process cannot be ignored when the corresponding arithmetics were implemented in practical hardware equipment, Wavelet Block Coding algorithm without link table based on block bit-length will have much more superiority, so this algorithm was employed in radar image compression system.In the last section of this dissertation, the hardware realization platform for radar image compression system was given in details, including its working principle, the introduction for main chips and the explanation for the implement of wavelet coding algorithm through the platform. Subsequently, the testing program loaded into PC and corresponding concrete program diagram, the main window for the testing program and parameter configuration window were presented as well. Finally, particular tests for radar image compression system were conducted according to some international testing standards set for radar image compression techniques in " Test Guidance for Shipborne Voyage Data Recorder" issued by CCS, the test result showed that the performance and its characteristics of this radar image compression system can reach the international standards entirely and it has a long time period working stability and reliability.更多还原