edge_gaijinsobel

探索图像恢复中出现的正则化方法及其理论基础，进一步发展Tikhonov正则化方法，得到图像恢复算法和工程实际中一些病态问题的解，进一步发展已有的数值解 法,(Exploration of image restoration emerged regularization method and its theoretical basis, the further development of Tikhonov regularization method, the image restoration algorithm and engineering solutions of some ill-posed problem, the further development of existing numerical method,)

说明：  基于radon正反变换的抛物线型多次波去除(Removal of parabolic multiple waves based on Radon transform)

核聚类算法KFCM源代码，对传统FCM算法的改进(Nuclear clustering algorithm KFCM source code, the traditional FCM Algorithm)

SIFT特征匹配算法是目前国内外特征点匹配研究领域的热点与难点，其匹配能力较强，可以处理两幅图像之间发生平移、旋转、仿射变换情况下的匹配问题，甚至在某种程度上对任意角度拍摄的图像也具备较为稳定的特征匹配能力。该算法目前外文资料较多，但中文方面的介绍较少.(SIFT feature matching algorithm is at home and abroad feature points matching hot area of ​ ​ research and difficulties in its matching capability that can handle between the two images, translation, rotation, the matching problem in the case of affine transformation, and even to some extent have a relatively stable feature of the ability to match images taken on any angle. The algorithm is more foreign materials, but the Chinese introduction less.)

视频异常行为检测，可以实现，使用matlab编程(Video abnormal behavior detection can be achieved using matlab programming)

对于iOS图像旋转，精选ios学习编程源码，很好的参考资料。(For iOS image rotation, selection of learning IOS programming source code, a good reference.)

Image_processing in C

G测地活动轮廓模型的代码，是基于图像的梯度信息进行处理的，属于水平集范畴(Measured the active contour model code is based on the gradient of the image information processing, belonging to the scope of the level set)

说明：  该课题为基于MATLAB平台的PCA的人脸识别图像考勤系统。传统的人脸识别都是直接人头和人头的对比，现实意义不大，这块都做烂了，不易通过毕设答辩，而且论文不好写。该课题识别原理为：从一副生活照中寻找到人脸，并且分割人脸图象，利用PCA算法进行降维，和库里图片进行对比，输出目标人脸以及记录识别的时间，可以统计人脸打卡的当月次数，从而统计出勤率。该课题还可以继续二次开发，做成库内外人脸的识别，如果是库外人脸则实现报警。GUI界面有清除数据功能，统计工资等功能。(This project is based on the MATLAB platform PCA face recognition image attendance system. The traditional face recognition is the comparison of direct head and head, which is of little practical significance. This part is not easy to pass the final defense, and the paper is not easy to write. The recognition principle of this subject is: find the face from a living picture, segment the face image, reduce the dimension by PCA algorithm, compare with the library picture, output the target face and record the recognition time, and then count the number of times in the month when the face is punched, so as to count the attendance rate. The subject can also be further developed to make face recognition inside and outside the database. If it is outside the database, the alarm can be realized. The GUI interface has the functions of data clearing and salary statistics.)

自适应多尺度复合形态滤波，对信号或图像实现滤波(Adaptive Multi-scale Compound Morphology Filter)