Arcgis制图规范符号库

基于信息熵的迭代法求最佳阈值，并附参考论文及KSW熵法求阈值的程序。两个程序经实验测试通过无误。(Entropy-based iterative method for optimal threshold, together with reference papers and KSW entropy threshold method procedures. Two experimental test procedures correctly.)

基于内容的图片搜索，基于java实现，图片搜索及图像相似度计算(Content-based image retrieval)

关于车牌字符识别的神经网络算法,才用的是三层神经网络(shengjinwangluosuanfa in car recognise )

利用MATLAB软件完成基于离散余弦变换（DCT）的可见水印算法，实现可见水印的嵌入(discussed the mathematical manipulation method of visible digital watermark and realized the visible digital watermarking inserting by completing the watermark algorithm，which based discrete cosine transformation (DCT) with the MATLAB software)

基于背景减除的单目标跟踪方法在Matlab中的实现，包含测试用avi文件(Background subtraction based on a single-target tracking method in Matlab, implementation, including test avi file)

说明：  susan算法是基于灰度的特征点获取方法，适用于图像的边缘提取(susan algorithm is based on the gray-scale method to obtain the feature points for image edge extraction)

说明：  基于MATLAB编程的菲涅耳离轴全息图的记录(Recording of Fresnel off-axis hologram)

sift详细编写,方向等参量，参数取舍详细(sift Detailed compilation of sift, direction and other parameters, detailed selection of parameters sift sift sift sift)

说明：  UNet最早发表在2015的MICCAI上，短短3年，引用量目前已经达到了4070，足以见得其影响力。而后成为大多做医疗影像语义分割任务的baseline，也启发了大量研究者去思考U型语义分割网络。而如今在自然影像理解方面，也有越来越多的语义分割和目标检测SOTA模型开始关注和使用U型结构，比如语义分割Discriminative Feature Network(DFN)(CVPR2018)，目标检测Feature Pyramid Networks for Object Detection(FPN)(CVPR 2017)等。(Its influence has reached 70% in 2015. Then it became the baseline that most of the medical image semantic segmentation tasks, and inspired a large number of researchers to think about the U-shaped semantic segmentation network. In the aspect of natural image understanding, more and more SOTA models of semantic segmentation and object detection begin to pay attention to and use U-shaped structure, such as semantic segmentation, discriminative feature network (DFN) (cvpr2018), feature pyramid networks for object detection (FPN) (CVPR 2017), etc.)

此程序是基于互相关的亚像素级图像匹配的代码，此种配准方法误差小，精度高。(This procedure is based on the cross-correlation of sub-pixel image matching of the code, such a registration method error and high accuracy.)