点云特征提取 详细 实用 经典的讲述过程及方法FeatureExtractionFromPointClouds

说明：  利用matlab进行编译，显示raw格式的裸数据(Compile with matlab and display raw data in raw format)

说明：  基于SegNet实现了车道线的识别。里面包含已经训练好的模型。(Lane line recognition based on SegNet contains the trained model.)

用画笔绘制圆角矩形，椭圆，饼形，并用不同颜色画刷进行填充。(Draw a rounded rectangle, ellipse, pie with a brush, and brush with a different color fill.)

自己画的一条简单的Bezier曲线，通过四个控制点（控制点可自行选取）(She painted a simple Bezier curve through four control points (control points can be selected as their own))

PCB板检测的大概流程如下：首先存储一个标准PCB板图像作为参考标准，然后将待检测的PCB板图像进行处理，比较与标准PCB图像的差异。根据差异的情况来判断缺陷类型。(The approximate process of PCB board detection is as follows: first, a standard PCB board image is stored as a reference standard, and then the PCB board image to be detected is processed to compare the difference between the standard PCB image and the standard image. The type of defect is judged according to the difference.)

基于混合高斯的运动目标检测程序已经相应的文章！非常值得借鉴！(Gaussian mixture-based moving object detection procedures have been the appropriate article! Well worth learning!)

Cohen类时频方法采用不同的核函数来抑制交叉项，本质上是在模糊域对信号进行低通滤波，其中典型的方法有Choi-Williams 分布和Cone-shaped 分布等。(The Cohen time frequency method uses different kernel functions to suppress the cross term, which is essentially a low pass filter for the signal in the fuzzy domain. The typical methods are Choi-Williams distribution and Cone-shaped distribution.)

说明：  将原图二值化处理后，对二值图像进行腐蚀处理去掉图中主体以外的噪点或小斑迹(After binarization of the original image, the binary image is corroded to remove the noise or small spots outside the main body of the image)

图形验证码识别引擎 V1.1 可以识别数字(Graphics Recognition Engine Code V1.1)

开发 API人脸检测实例教程，应用程序编程源码，很好的参考资料。(Tutorial examples of the development of API face detection, application programming source code, a good reference.)