Background modeling post_processing Common OPENCV functions (angry)

1.saturate_cast<uchar>, is to convert the data into 8bit 0~255 interval, negative value becomes 0, greater than 255 becomes 255. If it is a floating-point data, it becomes the round nearest integer

2.cv::mat::p tr<float> (y) [X];cv::mat::p tr<float> (y) is a pointer to the corresponding row.

The process of initializing 3.cv::mat::zeros () is y*x*c, remembering that it is not x*y*c.

4.canny for edge detection, the type of the image must be cv_8u, or the compilation will not pass.

5.imshow to visualize the matrix must be an integer to be able to AO, if it is floating point, not care you, direct white screen.

Background modeling post_processing Common OPENCV functions (angry)