curves-movement-by-level-set-method

图像检测的算法，Faster R-CNN算法，先对整张图像进行卷积计算，然后通过感兴趣区域池化层(RoI Pooling Layer)将选择性搜索算法推荐出来的候选区域和卷积网络计算出的特征映射图进行融合，得到候选区域对应的特征矢量，这种共享卷积计算的操作极大地减少了卷积计算的次数。而且这些特征矢量的维度统一，方便后续的分类工作。通过感兴趣区域池化层处理卷积特征，并将得到的特征送往两个并行计算任务进行训练，分类和定位回归。通过这些方法和改进的框架，Fast R-CNN 用更短的训练和测试时长，取得了比 R-CNN 更好的效果(Faster R-CNN algorithm first convolutes the whole image, then fuses the candidate regions recommended by the selective search algorithm and the feature mapping maps calculated by the convolution network through the RoI Pooling Layer to get the corresponding feature vectors of the candidate regions, which greatly reduces the number of convolution calculations. Moreover, the dimension of these feature vectors is unified, which facilitates the subsequent classification work. The convolution feature is processed by the pooling layer of the region of interest, and the obtained feature is sent to two parallel computing tasks for training, classification and positioning regression. Through these methods and improved framework, Fast R-CNN uses shorter training and testing time and achieves better results than R-CNN.)

关于阈值处理的一部分参考文献，对于图像的二值化等很有帮助，是经过精心选择的参考文件。(References on the part of the processing threshold for image binarization helpful, is carefully selected reference files.)

hausdorff距离的一种改进算法，是我自己编写的，用于提取焦点之后的图像配准(hausdorff distance of an improved algorithm, which I have written for the extraction of the focus after the image registration)

说明：  小波去噪 matlab小波去噪详解超全超全有程序,小波去噪，硬阈值软阈值(Wavelet denoising Matlab wavelet de-noising details super full super full program, wavelet de-noising, hard threshold soft threshold)

说明：  基于开源计算机视觉库OpenCV的图像处理基于开源计算机视觉库OpenCV的图像处理(Based on open source computer vision library OpenCV image processing based on open source computer vision library OpenCV image processing)

一个很强大的图形计算处理软件，包括图像聚类，矩阵计算等(Scotch is a software package for graph and mesh/hypergraph partitioning, graph clustering, and sparse matrix ordering )

WMIL跟踪算法的实现,代码注释全功能强大(WMIL tracking algorithm, code comments all powerful)

车辆牌照识别，比较简单的识别系统，可能由于数据库的不足，导致其还存在一些误差(Vehicle license plate recognition, a relatively simple recognition system, may be due to the lack of database, resulting in some errors.)

绣花问题是连续型的填充，在不同区域的缝制过程中，缝针不能提针，因此不能用一般的离散型区 域填充方法处理。本文运用几何计算和图论理论提出了一个绣花缝针轨迹自动生成算法：先对轮廓走向进行 定义，通过轮廓铅垂方向的局部极值点的分割线将图案从上向下进行分割，采用交点的特征值比较彻底的解 决了分割时的重点问题，将图案区域准确的划分成缝针能一次完成的若干个节点。在此基础上根据节点的邻 接关系建立节点的邻接“图”，通过图论中半哈密尔顿路径或深度遍历方法找到节点的遍历（缝制）序列和每 个节点的缝制方向，最后可对设定的起点、终点和缝针间隔的条件自动提供缝针的走向轨迹。(In the process of embroidering, needle can not hang up for changing to a different area. Therefore, Embroider problem could be abstract as a continuous area-filling work. Considering disperse area-filling methods can not process this problem well. Based on geometry computing and graph theory, an algorithm of finding path of embroidering needle are proposed in this paper. First, outline orientations are defined, and finding all local extreme points of inner outline on their gravitational orientation to build sectioning lines so as to divide the picture. By using intersection point character the overlap point problem was solved perfectly. Therefore, the picture is divided into some nodes which can be finished alone. Then, based on connections of these nodes, an adjacency graph of nodes was built. Using half-Hamilton path or depth-first search method, both embroidering sequence and direction of these nodes could be got from the graph. Finally, from a defined start-point, end-point)

很有用的彩色图像分割方法，经本人实验，绝对可靠(Very useful for color image segmentation method, by my experiment, absolutely reliable)