bp_erf_log

编程实现二维8*8图像块的DFT，DCT变换 输入一副RGB图像，将其转化为YCbCr颜色空间，然后对Y分量做分块DFT变换，保存为灰度图像，统计计算时间 编程实现8*8块DCT变换,然后分别显示，当使用64个DCT系数里面的一些系数进行重构，比较其质量；同时熟悉采用FFT来实现DCT变换的思想，并显示2维8*8DCT变换的基函数图像(Programming the DFT, DCT transform of two dimensional 8*8 image block Input a pair of RGB images, transform it into YCbCr color space, then divide the Y component into block DFT transform, save as gray image, and calculate the calculation time The 8*8 block DCT transform is implemented by programming, and then it is shown that when using some coefficients in 64 DCT coefficients, the quality is reconstructed. Meanwhile, the idea of DCT transformation is familiar with FFT, and the base function image of the 2 dimensional 8*8DCT transform is displayed.)

经典亚像素边缘检测算法Zernike算法程序，编译过可以运行！！(Classical active contour the snake model program, you can run the compiled! !)

我自己做的一个bmp文件进行24位真彩转256灰度和二进制的程序，并能用sobel算法，识别物体的边界。 (I myself make a bmp file for 24-bit true color to 256 gray-scale and binary procedures and sobel algorithm can be used to identify object boundaries.)

matrix 基本功能集合,认真研究会有收获()

用Matlab来实现PCA，并分别求出图像在第一、二、三主分量上的投影。(Using Matlab to implement PCA, and images were obtained in the first, second and third principal component on the projection.)

Solutions to problems in the field of digital image processing generally require extensive experimental work involving software simulation and testing with large sets of sample images.

说明：  QC-LDPC的BP译码算法，MATLAB实现，简单易懂，已成功(Various encoding and decoding methods of LDPC)

说明：  灰度不均匀图像处理，具有优良效果，可以使用，效果较好(Uneven grayscale image processing, with excellent results, can be used, with better results)

指纹图像增强，gabor滤波器，包括形态学分割，形态学细化，加窗短时傅里叶变换增强，方向图估计及平滑，纹线频率估计及平滑(Fingerprint image enhancement, gabor filters, normalization, equalization, morphological segmentation, morphological thinning, windowed short time Fourier transform enhanced estimates and smooth pattern, ridge frequency estimation and smoothing)

圆形，孔型工件的尺寸测量，主要采用canny算子检测边缘，并进行亚像素细分，还有GUI，测量精度在0.1mm(Circular hole workpiece size measurement using canny operator edge detection and subpixel, as well as GUI, measurement accuracy of 0.1mm)