ColorImageSegmentation

本程序的细化算法要应用于含线条的黑白二值图像才可以看出细化效果。(This procedure should be applied to the thinning algorithm with two lines of black-and-white image before thinning effect can be seen.)

这是一种彩色图像去雾算法，可以识别物体的轮廓，但是同时也会加强这些轮廓，虽然去雾效果不错，但是会加强轮廓！(This is a color image to fog algorithm can identify the contours of the object, but will also enhance the outlines, although to fog effect is good, but will strengthen the profile!)

说明：  利用Hough变换实现半径未知的圆形物体的检测(Use Hough transform to achieve a circular radius of an unknown object detection)

对一幅图像进行DCT与DFT变换，分析变换后的高低频分量系数(DCT and DFT on an image, analysis of high frequency components of the transformed coefficient)

应用Intel® JPEG Library v1.5处理图像的例子(Using Intel® JPEG Library v1.5 to process image.)

Retinex算法对较暗区域的图像处 理有明显效果,改进的Retinex方法则是基于经典Retinex,对具有较高亮度的遥感图像通过图像变 换方法达到去除模糊的目的。 (Retinex algorithm for image processing on the dark region was effective to improve the Retinex method is based on the classic Retinex, on remote sensing images with high brightness achieved by an image transformation method aims to remove ambiguity.)

激光三维点云的处理源码，可以压缩点云，配准点云(Three-dimensional laser point cloud processing source code, can be compressed point cloud, point cloud registration)

提出了一种基于非下采样Contourlet 变换的自适应图像去噪方法 首先对噪声图像进行非下采样Contourlet 变换，得到各 个尺度各个方向子带的系数，再根据该系数的能量自适应地调整去噪阈值 实验表明，与Contourlet 多尺度阈值去噪 Contourlet 自 适应阈值去噪相比，该方法在保留图像边缘细节的同时，能提高图像的PSNR 值，减少了Gibbs 现象(A new adaptive method of image de-noising based on NonSubsampled Contourlet Transform （NSCT）is presented.Firstly， the noisy image is decomposed into a set of multiscale and multidirectional frequency subbands by NSCT， and then the threshold is adapted automatically according to the energy of subband coefficients.Compared with the multi-scale threshold using contourlet transform and using adaptive threshold based on contourlet transform， the simulation results show that the performance of this method is superior in PSNR， meanwhile it can maintain the image edge and reduce the Gibbs phenomen)

用差分盒维数处理图像，通过对图像分块，然后每一块来计算该区域的分维数，将其作为该区域所有像素上每一像素的分维数。然后，增强分维数分布图效果，然后取阈值，最后，给人造目标加框。(Differential box counting process the image, the image block, and then every one to calculate the fractal dimension of the region, as all the pixels in the region of each pixel on the fractal dimension. Then enhanced fractal dimension distribution effect, and then take the threshold, and finally, to the man-made objects framed.)

用PHD滤波的方法实现多目标跟踪，及对PHD滤波性能的检测。(PHD filter method using multi-target tracking, and performance testing for the PHD filter.)