图像处理中的一些常用代码Image-processing-code(4)

本文主要介绍遗传算法的基本理论，叙述遗传算法在图像增强的的主要应用，即将原始图像变得更加清晰，特征变得更加明显。 现今关于图像增强的算法有很多，而这些算法大多是基于退化函数或者点扩展函数的知识进行图像处理的。当图像出现模糊或噪声影响大时，设计出的图像清晰化的效果肯定不够理想，因此有必要对图像进行增强处理。于是，可利用到遗传算法这种成熟稳定的仿生物进化的全局寻优算法，进行图像增强，由于遗传算法控制参数少、自适应度高，将选择该方法对图像退化分别进行不同的清晰化处理. 这样既增强了图像的对比度,又克服了传统直方图均衡化处理所造成的灰度级损失等缺点。 通过实验去表明，遗传算法从全局寻优的角度出发，能够较为精确地估计退化系统函数。和传统的线性增强、直方图均衡进行比较。实验结果表明该方法能改善原图像视觉效果，便于之后的图像分析。 (This paper mainly introduces the basic theory of genetic algorithm, genetic algorithm (ga) in the main application of image enhancement, the original image is more clear, characteristic becomes more obvious. Today about the image enhancement algorithm are many, and most of these algorithms is based on the degradation function or the knowledge of the point spread function of image processing. When the image appears fuzzy or noise influence, designed the image clarity of effect must not ideal, therefore it is necessary to enhance the image. Then, genetic algorithm is available to the mature and stable evolution of the global optimization algorithm, for image enhancement, due to less genetic algorithm control parameters, since the high fitness, will choose the method of image degradation, respectively, to manipulate the different motivation. It can enhance the image contrast, and overcome the traditional histogram equalization processing grayscale loss caused by the faults. Throug)

将一副图像进行rgb分离，并将R G B 的各个分量显示出来。(the sepration of rgb and show the sepration )

基于MM-GMPHD算法多目标跟踪，对目标数目，观测量进行估计，共有五个不同时刻产生的目标，(multi-target tracking )

KLT point match:Detection and Tracking of Point Features.

利用蝶形算法类似于FFT，利用先行再列实现二维DCT。速度很快。包含反DCT的代码。(similar to the use of butterfly FFT algorithm, the first re-use of 2D DCT out. Very fast. Contain anti- DCT code.)

运用Perona-Malik各项异性扩散模型对二维图像进行去噪处理。该方法在有效的抑制噪声的同时能够保护图像的边缘。(The use of Perona-Malik Anisotropic Diffusion Model for the treatment of two-dimensional image denoising. The method in the effective suppression of noise at the same time to protect the edge of the image.)

等效视数表示遥感图像滤波后的滤波效果，这个程序可以计算等效视数。(Equivalent number of said filtering effect, this program can calculate the equivalent number of remote sensing images filtered.)

图像交互是分割 graph cut,实现交互式分割，并且可以连续分割(Interactive image is split graph cut, to achieve interactive segmentation, and can be continuously split)

扩展卡尔曼滤波的基础代码，有详细的注释，结合卡尔曼滤波的基本原理，就能很好的理解卡尔曼滤波过程了。(Extended Kalman filter based code , there are detailed notes , combined with the basic principles of Kalman filtering , we can well understand the process of Kalman filtering )

sift算法，特征点的提取与匹配，图像融合，图像拼接。测试过，效果不错，特征点提取较稳定。(Sift u7B97 u6CD2 uFF0C u7279 u5F1 u70B9 u7684 u63D0 u53D6 u4E0E u5339 u914D uFF0C u56FE u50CF u878D u5408 uFF0C u56FE u50CF u62FC u63A5 u3002 u6D4B U8BD5 u8FC3 uFF0C u6548 u679C u4E0D u9519 uFF0C u7279 u5F81 u70B9 u63D0 u53D6 u8F83 u7A33 u5B9A u3002)