语义分割

说明：  分子图像处理的软件，但没有数据库，具体操作看操作说明(Molecular image processing software, but there is no database, see the specific operation instructions)

填充区域图像，设置填充区域，填出图像的颜色(Filled regional images, set fill area, fill in the color of the image)

设置位置断点跟踪程序，VC++精选编程源码，很好的参考资料。(Set breakpoint location tracking procedures, VC++ programming source code, a good reference.)

说明：  ORB-SLAM 基本延续了 PTAM 的算法框架,但对框架中的大部分组件都做了改进, 归纳起来主要有 4 点: ORB-SLAM 选用了 ORB 特征, 基于 ORB 描述量的特征匹配和重定位, 都比 PTAM 具有更好的视角不变性。此外, 新增三维点的特征匹配效率更高, 因此能更及时地扩展场景。扩展场景及时与否决定了后续帧是否能稳定跟踪。ORBSLAM 加入了循环回路的检测和闭合机制, 以消除误差累积。系统采用与重定位相同的方法来检测回路(匹配回路两侧关键帧上的公共点), 通过方位图 (Pose Graph) 优化来闭合回路。PTAM 需要用户指定 2 帧来初始化系统, 2 帧间既要有足够的公共点, 又要有足够的平移量. 平移运动为这些公共点提供视差 (Parallax) , 只有足够的视差才能三角化出精确的三维位置。ORB-SLAM 通过检测视差来自动选择初始化的 2 帧。PTAM 扩展场景时也要求新加入的关键帧提供足够的视差, 导致场景往往难以扩展. ORB-SLAM 采用一种更鲁棒的关键帧和三维点的选择机制——先用宽松的判断条件尽可能及时地加入新的关键帧和三维点, 以保证后续帧的鲁棒跟踪; 再用严格的判断条件删除冗余的关键帧和不稳定的三维点，以保证 BA 的效率和精度。(Orb-slam basically continues the algorithm framework of PTAM, but most of the components in the framework have been improved, which mainly includes four points:)

小波去噪,实现阈值小波去噪，并且含有frost去噪，并且输出信噪比(Wavelet denoising, to achieve the threshold wavelet denoising, denoising and contains frost, and the output signal to noise ratio)

人脸检测 通过肤色建立模型，和计算欧拉数开运算等一系列处理可以检测出同一张图片中的多个人脸！效果不错哦(face detection through color model building, and calculation of Euler number opened a series of processing operations can be detected with a picture of the personal face! Oh, good results)

IEEE:数字水印技术引论，其中将达哦水印技术的引出、算法发展、引用等(IEEE: primer on digital watermarking technology, which will lead to Oh watermark technology, algorithm development, quoted, etc.)

将肿瘤图像中的肿瘤轮廓提取，将轮廓的矩阵输入到函数中，得到肿瘤的分维结果。(The tumor contour is extracted from the tumor image, and the matrix of the contour is input into the function to obtain the fractal dimension of the tumor.)

将某一8位的raw格式图转换为16位的raw格式图 并可输出raw格式图的灰度值至相应文件（.txt文件） 生成的新图用PS打开时与原图基本一致.(It can converted an 8-bit raw format, to an 16-bit raw format and can output raw format diagram of the gray value to the corresponding map file (. Txt files) .The new plan to open with PS are basically consistent with the Original.)

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.)