[波变换] JPEG2000_9_7_002.pdf
基于实数的二进制表示法，把CDF(Cohen，Daubechies and Feauveau)9／7双正交小波基的提升系数化为二进制，采用简单的移位一加操作代替结构复杂的浮点乘法器，从而实现了JPEG2000中9／7离散小波变换的定点计算．相对于浮点计算法，移位一加操作最大的优点是计算简单，特别易于超大规模集成电路实现，因而使硬件实时处理图像信号成为可能．实验仿真结果表明：在低压缩比的情况下，用移位一加操作重构的图像，其峰值信噪比(PSNR)只比浮点法低0．10 dB，当压缩比增大时，其PSNR值略好于浮点法．
关键词： 离散小波变换；定点计算；浮点计算法；提升；移位一加操作
Based on the real number of binary notation, the CDF (Cohen, Daubechies and Feauveau) 9/7 Biorthogonal wavelet enhancement factor into the binary, using a simple shift operation in place of one plus the complex structure of floating-point multiplier, in order to achieve the JPEG2000 9/7 discrete wavelet transform the fixed-point calculation. Compared with the floating-point calculations, shift operation of one plus the biggest advantages is the calculation is simple, especially vulnerable to the realization of ultra-large-scale integrated circuits, making real-time image processing hardware signal possible (2009-01-19, C/C++, 221KB, 下载25次)
[波变换] DSP_JPEG2000_001.pdf
本文介绍了JPEG2000的静态图像压缩技术在TI TMS320DM642 DSP上的实现过程；对压缩技术核心小波算法结构进行改进，对硬件特点的程序结构、具体代码及总线读写等方面的改进作了详细描述．改进后的软件系统，对于1 024×768像素大小的图像、数据存储空间仅需6 KB，使得在资源有限的DSP芯片上能够运行JPEG2000图像压缩技术；编码速度相对于未改进的提高了l2～ 2O倍，可迭到3帧／s；压缩比为O．5 bit／p时，峰值信噪比可达到3O．
This paper introduces the JPEG2000 still image compression technology in the TI TMS320DM642 DSP Implementation process on wavelet compression technology is the core algorithm to improve the structure of the hardware features of the process structure, read and write the specific code and bus improvements, such as was described in detail . The improved software system, for 1 024 × 768 pixel image size, data storage space only 6 KB, allows limited resources to run the DSP chip JPEG2000 image compression technology encoding speed is not improved relative to the improvement of the l2 ~ 2O-fold, again and again to be 3/s compression ratio for O. 5 bit/p, the peak signal to noise ratio can be achieved 3O. (2009-01-19, C/C++, 238KB, 下载35次)
[波变换] MyDwt2D
对图像进行二维离散小波变换, 变换级数大于等于3级,然后统计系数中0的个数（百分比表示），并进行重构, 最后计算重构图像的峰值信噪比(PSNR).
里面还附上了我写的实验报告，对大家理解Mallat算法和图像压缩很有帮助
Image two-dimensional discrete wavelet transform, transform series greater than or equal to 3, and then statistics of the number of coefficients in 0 (expressed as a percentage), and reconstruction, the final calculation of reconstructed image of the peak signal to noise ratio (PSNR). which also includes the experimental report I wrote to all of us understand the Mallat algorithm and image compression helpful (2008-11-21, C/C++, 963KB, 下载126次)