【作者】 杨扬；

【导师】 邓家先；

【作者基本信息】 海南大学 ， 信号与信息处理， 2012， 硕士

【摘要】 随着现代图像传感器的空间分辨率不断提高以及航天技术的快速发展,图像压缩已成为一种基本技术。当前,各国航天局从太空中获取的高品质图像和高端医疗设备产生的医学图像,如BMP、TIFF或任何未压缩格式的高清图像会占用大量的存储空间,对数据存储容量和传输带宽带来了巨大挑战。在图像质量要求很高的遥感、医疗图像等特殊领域,有时候需要对图像数据进行无损或者近无损压缩,以得到高质量的重建图像。无损压缩标准JPEG-LS能够进行无损图像压缩,或者能够实现误差可控的近无损数据压缩。JPEG-LS的核心算法是一种利用上下文模型和Huffman编码的LOCO-I(Low Complexity Lossless Compression for Images)算法。该算法与JPEG2000和JPEG算法相比较,主要在于利用了Golomb行程编码,引入误差可控的近无损图像压缩,在近无损图像压缩方面更具优势,由于其算法复杂度相对较低,更适合FPGA的硬件实现。然而,Xilinx公司提供可用的JPEG-LS压缩算法的商用IP核的价格昂贵,且完全封装化,在产品开发中可扩展性差。本文研究了JPEG-LS压缩算法的编解码原理,对核心算法的上下文建模、Golomb-Rice编码、游程和常规模式的具体流程进行了详细分析,并在PC机上采用C语言对算法的编解码进行建模,且通过了大量特殊图像数据的测试,验证了硬件实现的可行性。在此基础上,进行解码器的系统设计,考虑到解码的实时性,以图像分块解码的方式,将原始图像分成8块,分别送入8路解码模块进行并行解码。整个设计采用VHDL硬件语言描述,通过功能仿真和静态时序分析,协同FPGA硬件开发板共同验证,解码器可正确实现遥感图像的解码。本解码系统在实时性和稳定性上都可满足遥感图像解码压缩的各项技术指标要求。更多还原

【Abstract】 Along with the improvement of modern image sensor’s spatial resolution and the rapid development of aerospace technology, image compression has become a basic technology. Currently, national space agency obtains images with high quality from space and medical images from high-end medical equipment. The obtained images such as BMP, TIFF or any uncompressed HD images will take a lot of storage space, this brought enormous challenge to the capacity of data storage and transmission bandwidth. In remote sensing, medic and some other special areas requiring high image quality, Sometimes need to lossless or nearly lossless compression to get the high quality of the reconstructed image of image data. Standard JPEG-LS lossless compression can be lossless image compression, or able to achieve error-controlled near lossless data compression.LOCO-I (Low Complexity Lossless Compression for Images) algorithm based on Huffman encoding and the context model are the core of JEPG-LS algorithm. Compared with JEPG2000and JEPG algorithm, this algorithm is characterized with the use of Golomb coding and the injection of error controllable in near-lossless image compression.making it more advantageous in the near-lossless compression and more suitable for hardware implementation of FPGA.But the IP core that the Xilinx company launched for JPEG-LS compression algorithm is expensive,complete packaged and has poor scalability in product development.This article studied the JPEG-LS compression algorithm encoding and decoding principle carefully, and exact analyzed the core algorithm of context modeling, Golomb-Rice coding, run-length and conventional patterns of specific process. In order to verify the hardware realization feasibility, we modeling the encoding and decoding algorithm using C language on a PC and test it with large image data. depend on above all, the system design of the decoder. Taking the real-time decoding into account, using image block decoding method, divides the original image into8blocks then respectively sent them into the8decoding module for parallel decoding.The entire design using the VHDL hardware description language, by functional simulation and static timing analysis, collaborative FPGA hardware development board to verify that this decoder can correctly decode the remote sensing image.The design can meet the technical requirements of remote sensing image decoding compressed.in the aspect of real-time performance and reliability.更多还原