CALCULATE TRAFFIC CONGESTION VIA CCTV

Capstone project in Handong University with Prof Yi

TRAFFIC CONGESTION CALCULATOR is for calculating traffic congestion through CCTV images. Firstly we find non-edge area, and then choose the biggest one in non-edge areas. Using that contour information especially scalar value, re-use biggest-contour's LAB value, and filter the image, then find the road. Furthermore, calculate the portion of objects on the road and then convert the values to traffic congestion.

How To Use?

RoadDetection.h is a header file.
In this program, we use OPENCV (open-source software).
Import that header file with cpp file, then just use functions!

1. Header Explanation :

----------  Finding Road part  -----------

1. Normalization : Normalize the matrix (Scalar value).
2. nonedge_area : Find non edge area. Supposed non edge area was road.
3. FindLargestArea : Find the largest area in non-edge areas.
4. roadFilter : Contract LAB values for filtering from non-edge-largest area.
5. roadFilter2 : Same with roadFilter, but used BGR only.
6. callBackFunc2 : For checking scalar values from picture where clicked.
7. FindRoad : Find road area using similar 'LAB & BGR' in largest non edge area. Combined all function above.
(If you want to find road, just use this function.)

----------  Count car  -----------
(Still evaluating)

-int ifLRline(Point2f A, Point2f B, Point2f P);
-bool intersection(Point2f o1, Point2f p1, Point2f o2, Point2f p2, Point2f &r);
-void OpticalFlow_Count(int Pnum, vector<uchar> status, int & Car_num, Mat& frame, Point2f & pass, vector<Point2f> after, vector<Point2f> Center, Point2f A, Point2f B);
-double dist(Point2f A, Point2f B);

----------  Machine-Learning for detect Car  -----------
(Learning is another part, this code is just for importing Haar-cascade)

1. detect_haarcascades: Input source & Input Cascade XML's path.

----------  Image Compare for rotate camera  -----------

1. DistHisto : Compare HSV histogram with distributed matrix. Not all part of Image (distribute, and compare). - Recommended -
2. calcEdgeDirection : Calculate pixel's direction, and prepared for using it for comparing images (Seems like HOG).
3. calcEdgeDirection2 : Upgraded version. Plus magnitude calculation. - Recommended - 
4. DistEdgeCompare : Compare 'Edge direction's histogram' with images. - still evaluating. maybe good for finding road-line -
5. DistEdgeCompare2 : Upgraded version. Pluse magnitude calculation.

-------- etc ---------
direct_area, nomalize, calcEdgeDirection3

-> is for testing.