Simple Computer vision project (good example for using openCV 3 on C++ style)
This project detects all threads in textile machine (and detects cases, when threads are broken or other object overlap threads)
For installing and compiling project you need to install:
- openCV with extra contrib modules (Compile and install tutorial for Ubuntu)
- g++ with c++11/14 standart supporting
This project uses next algorithms:
Haar's cascadefor detecting threads and wheels on textile machine (dataset for training in./cascadeHaardirectory)
void detect(cv::Mat & frame, int neighbors, cv::Size min_size, cv::Size max_size);
cv::Mat & frame - captured frame from video or web-cam (and transformed to grey binary image)
int neighbors - normal overlaping coefficient (for `cv::CascadeClassifier::detectMultiScale` openCV method)
cv::Size min_size - minimal rectangle size for detecting (for `cv::CascadeClassifier::detectMultiScale` openCV method)
cv::Size max_size - maximal rectangle size for detecting (for `cv::CascadeClassifier::detectMultiScale` openCV method)BackgroundSubstractorMOGfor detecting other object, that overlaps threads and textile machine (usescv::BackgroundSubtractorfrom OpenCV contrib modules)
bool detectOther(cv::Mat & frame);
cv::Mat & frame - captured frame tranformed to binary image
returns true, when overlapped object detected> cd path/to/project
> make
> ./detector path/to/video (if no arg - webcam turn on)