wenlichengxu

图像处理中的深度学习，主要是原理的讲解，比较易懂,对于。(Image processing in deep learning, mainly to explain the principle, relatively easy to understand.)

给予分水岭的图像分割，包括源代码、测试图、实验报告(Given watershed image segmentation, including source code, test plans, test reports)

本例程序包含OMP、SP、IRLS、CoSaMP等重构算法，比较性能优劣。(The example application consists of OMP, SP, IRLS, CoSaMP reconstruction algorithm, and compare performance advantages and disadvantages. )

强大的小波包分析程序，将信号分频分析，在感兴趣的频率域内抽取信号信息。(Powerful wavelet packet analysis procedures, the signal of frequency analysis, frequency domain of interest extracted signal information.)

根据D. Comaniciu, P. Meer: Mean Shift: A robust approach toward feature space analysis 以及 C. Christoudias, B. Georgescu, P. Meer: Synergism in low level vision.这两篇文献提供的方法编写的图像分割代码，作者是 Chris M. Christoudias, Bogdan Georgescu，代码经我看了后加了丰富的中文注释，希望可以给各位带来阅读上的方便。 基于meanshift的聚类分割方法包括滤波、区域融合等操作，通过调整sigma和sigmar来调整分割效果。(According to D. Comaniciu, P. Meer: Mean Shift: A robust approach toward feature space analysis, and C. Christoudias, B. Georgescu, P. Meer: Synergism in low level vision. These two documents prepared by the methods provided by image segmentation code , the author is Chris M. Christoudias, Bogdan Georgescu, after I read the code, add a rich Chinese notes, hoping to bring you the convenience of reading. Segmentation method based on clustering meanshift including filtering, regional integration and other operations, and by adjusting the sigma sigmar to adjust segmentation results.)

说明：  SAR影像单视数据转为多视数据，数据间相互转化，快捷处理SAR数据(Single-look SAR image data into a more visual data, data inter-conversion, fast processing SAR data)

说明：  采用积分图算法的归一化互相关匹配，大大减少了计算时间。(Normalized cross correlation matching)

压缩感知图像重构的算法，直接观测原始信号（二维），最后重建(Compressed sensing image reconstruction algorithm, direct observation of the original signal (two-dimensional), and finally reconstruction)

利用傅立叶变换和相位相关法来检测图象的区域复制伪造(Use of Fourier transform and phase correlation method to detect forged images of the regional replication)

该程序是对原图像进行梯度加掩模的三次分水岭分割(That the procedure was in the original image mask plus three gradient watershed segmentation)