heat-source

对交通标志进行检测和识别（本文件仅针对红色交通标志进行识别）(Identification and identification of traffic signs (this document is only for identification of red traffic signs).)

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.)

说明：  视频追踪的框架算法，实现了基于meanshift和feature based的跟踪算法(Framework for video tracking algorithm meanshift based and feature based tracking algorithm)

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.)

3D 点云建筑物分割代码3D 点云建筑物分割代码3D 点云建筑物分割代码(3D point cloud, buildign segmentation3D point cloud, buildign segmentation3D point cloud, buildign segmentation)

波变换的图像配准与融合算法的研究，主要描述了图像的各种配准方法和基本的融合方法(Wavelet transform algorithm for image registration and fusion research, will describe a variety of image registration methods, and basic integration methods)

遥感影像道路网提取，遥感影像道路网提取，遥感影像道路网提取，(Remote sensing images road network extraction, remote sensing images road network extraction, remote sensing images road network extraction,)

带运动估计的时域滤波，一种基于小波的运动估计方法，可以有效的去除视频流的帧间冗余。与传统的帧参考模式的运动估计相比，没有错误漂移现象。内含编码器和解码器。(Belt Movement estimated time domain filtering, a wavelet-based motion estimation methods, can effectively remove the inter-frame video stream redundancy. With the traditional frame of reference than the motion estimation, no errors drift phenomenon. Intron encoder and decoder.)

说明：  PCL是一个类似于OpenCV的开源库，提供了很多三维点云的处理功能，其中就包括点云拼接。在三维扫描项目中，需要利用点云拼接方法将多次扫描得到的点云数据拼合成一个整体。在本例中，用一幅包含人脸的深度图像去匹配之前获得的人脸模板，这将使我们能够确定人脸在场景中的位置和方向。(PCL is an open source library similar to OpenCV, which provides a lot of processing functions for three-dimensional point clouds, including point cloud mosaic. In the three-dimensional scanning project, it is necessary to use point cloud mosaic method to combine the point cloud data obtained from multiple scans into a whole. In this example, a depth image containing face is used to match the face template obtained before, which will enable us to determine the position and direction of the face in the scene.)

对打开的图像，可以提取其纹理特征，程序运行无误(Right to open the image, you can extract the texture features, the program is running correctly)