基于医学图象的压缩方法研究

【作者】 陆峰；

【导师】 郭振民；

【作者基本信息】 扬州大学 ， 计算机应用技术， 2003， 硕士

【摘要】 随着图象处理技术和计算机技术的不断发展，数字成象技术在医学中的应用日益广泛，如在X射线、计算机断层技术(CT)、核磁共振技术(MRI)、计算机放射成象技术(CR)、血管数字减影技术(DSA)、超声图象(US)等中的应用；另一方面以计算机网络为基础的图象存储传输系统PACS(Picture Arc(?) and Communication System)及其应用也在不断地发展，所产生医学图象的数据量是相当巨大的。本文针对“在对医学图象进行压缩编码时不允许丢失有用的细节信息”这一特点，围绕PACS中的研究重点“图象的编码压缩”展开讨论和研究。 本文通过对小波变换理论及SPIHT(Set Partitioning In Hierarchical Trees)等算法的分析研究，提出了一种基于第二代小波变换、SPIHT算法与自适应算术编码相结合的图像压缩的编／解码方案，并用VC++6.0进行了编程模拟。接着，在对原始图象的统计特性及直方图特性进行分析比较后，进一步改进了我们的编／解码方案。对于所用的SPIHT算法，在具体编程实现时，我们做了相应的改进，使SPIHT算法的运行时间大大降低。并通过在自适应算术编码前，重组SPIHT算法后的码流，使本方案中的图象压缩比有进一步的提高。最后，我们对CDF(2，2)和FBI 9-7小波基的压缩性能进行了分析和比较，验证了：到目前为止，还没有一种小波基在对任何一幅图象进行压缩时，能在各个方面都体现出优势；并通过对SPIHT算法中取不同阈值时自适应算术编码效率的分析及不同图象之间SPIHT算法效率和自适应算术编码的效率分析，得出“SPIHT算法的编码效率与二代小波变换后小波系数为零的个数有最直接的关系，零的个数越多，SPIHT算法压缩效果越好”这一结论。在具体实现时，本文的方案获得了良好的无损或近无损压缩效果。研究表明这一方法充分利用了医学图象的统计特征，对于医学图像具有较高的无损压缩率，压缩的空间和时间复杂度低，利用分层存储或传输可以获得满意的压缩比和PSNR，达到“视觉无损、数据近无损”的目的。对于典型的医学图像(X线图像、CT图像、MRI图象等)的存储、传输等和检索应用，本方案是十分有效的。更多还原

【Abstract】 With the development of image processing and computer technologies, the digital image technology is increasing widely applied in the medical image areas, such as X-ray, CT, MRI, CR, DSA and US. At the other hand, PACS(Picture Archival and Communication System) and its application which basing on Computer network develop unceasing, and the amount of image data is vast. In the light of the characterisuc "May not lose any detail when compressing and encode with the medical image", this thesis revolves round the image compression and encoding to do the research.In this thesis, we present a image encode/decode compression system that synthesizes the lifting wavelet transform(SGWT) , SPIHT(Set Partitioning In Hierarchical Trees) algorithm and adaptive arithmetic coding., and simulate with the programs of VC++6.0. After comparing the original image statistical characteristic and histogram feature, we modify the encode/decode scheme. We amend the SPIHT algorithm in our scheme to decrease the run time, and increase the image compress ratio by reconstructing the code bits before the adaptive arithmetic codes. Based on analyzing and comparing the constringent performance of CDF(2,2) and FBI9-7 wavelet, we verify that "there is no wavelet that consistently performs better than all the other wavelets on all the test images." We analyze the adaptive arithmetic coding efficiency when choosing different threshold values, and different images in SPIHT algorithm Efficiency, and draw a conclusion that "There are straight forward relations between the coding efficiency of SPIHT algorithm and the number of zero in wavelet factor after lifting wavelet transform. The more large the number of zero are, the more compression effect of SPIHT algorithm is good." This system has a better effect on lossless and little lossy compression. The study indicates that the method make full use of the medical image statistical characteristics, the method achieves upper compress ratio, lesser spatial and time complex. Using progressive transmission will attain the appointed compress ratio and PSNR. For typical medical images (X-ray image, CT image, MRI image) storage, transfer and search, this system is sufficiently valid.更多还原