vfv1.0

visionpro字符识别

gray image to color image conversion

利用随机霍夫变换进行圆的检测，matlab编写(Randomized Hough transform for circle detection, Matlab prepared)

说明：  基于MATLAB实现传统图像去噪算法（均值滤波、中值滤波、非局部均值滤波NLM、三维块匹配滤波BM3D）(The traditional image denoising algorithms (mean filter, median filter, NLM, 3D block matching filter BM3D) and dncnn image denoising algorithm based on deep convolution neural network are realized based on MATLAB.)

传统的几种图像增强与去噪方法，包括直方图处理，灰度变换，滤波去噪等几类。细分为直方图均衡化，直方图规定化，灰度线性变换，中值滤波，维纳滤波，同态滤波(The traditional methods of image enhancement and denoising include histogram processing, grayscale transformation, filtering and denoising. Subdivided into histogram equalization, histogram specification, gray level linear transformation, median filtering, Wiener filtering, homomorphic filtering)

isomap matlab的实现源程序，使用时，需要添加主程序。(isomap Matlab implementation source program, when used, need to add the main program.)

说明：  解决散斑图像的重建问题,主要是使用了相位恢复算法(to study the picture problem)

利用形态学分水岭算法对图像进行分割，对细胞图像能达到较好的分割效果(Using morphological watershed algorithm for image segmentation, the cell image segmentation can achieve better results)

asift算法进行图像配准，具有较好的结果，调试完成，可用(image registration, asift algorithm has a better result, debugging is complete, the available)

图像修复是对图像中破损区域进行信息填充，以减少图像破损所带来的信息损失的过程。 传统的图像修复方法需要依赖图像的具体结构来制定相应的修复方法，压缩感知理论的提出，使得可以利 用信号的稀疏性来对图像进行修复。基于K 奇异值分解（KSVD）与形态学成分分析（MCA，Morphological Component Analysis）的图像修复方法首先采用形态学成分分析方法对破损图像进行特征分析，将其分解 为结构部分和纹理部分；然后基于学习型字典KSVD分别对这两部分进行过完备字典训练；最后利用训练得 到的字典实现对破损图像的修复。相比于传统的图像修复方法，该方法具有适应性强、修复效果好等优点(Image inpainting is to fill the missing data in corrupted images and thus to reduce the information loss of damaged image. Traditional inpainting algorithms are dependent on specific structure of target images compressive sensing theory makes is possible to realized image inpainting with signal sparsity. This paper proposes a novel inpainting algorithm based on KSVD and MCA algorithm, which first decomposes the image into texture part and structure part, and then trains the two dictionaries for these two parts with KSVD and reconstructs the original image with these two trained dictionaries. Experiment indicates that the proposed algorithm is of better adaptability and performance as compared with traditional algorithms.)