Similar to the last Hough Line detection, the following example illustrates the use of the new LineSegmentDetector class in the Imgproc module. Instead of using the new command, we have to use the Imgproc.createLineSegmentDetector() function to create a new instance of the class.
In addition to the Hough circle detection, this example works on the Hough line segment detection. It inputs the live webcam image; converts it into greyscale; applies a medianBlur filter; processes the Canny edge detection. The Imgproc.HoughLinesP() function will finally single out the line segments into a Mat – lines in our example codes.
The example explores the Hough Circle detection in the Imgproc module. It starts with a greyscale copy of the live webcam image with an application of a blur filter, in this case, a medianBlur.
This example continues from the last post to compute the optical flow between 2 greyscale images by using the calcOpticalFlowPyrLK() function in the Video module. The new position of the pixels tracked will be delivered in a MatOfPoint2f object. By using the last and current position of the feature points, we can plot the path of the pixel movements. Furthermore, we can use such information for interactive or generative drawings, found in my artwork, Movement in Time.
The coming example will be the sparse optical flow. Before that, we first work on the 2D feature points tracking. The function goodFeaturesToTrack() belongs to the Imgproc module. It takes in a greyscale image and identifies the feature points (corners) as a matrix of point, MatOfPoint. The sample code here uses the feature points to render a live graphics of the webcam image.
This example continues to explore the Video module in OpenCV. It uses one of the BackgroundSubtractors, the BackgroundSubtractorKNN. It learns the motion in front of the camera and treats the stationary scene as background.
In the code, the important command is
bkg.apply(frame, fgmask); |
The subtractor object bkg takes in the latest frame and generates a foreground mask, fgmask. We can use the foreground mask to single out the foreground moving object.
Here are a number of screen shots from the testing. The background is the planet Earth image from NASA.
In this example, we move on to the Video module of OpenCV 3.0.0. The first function we test is the Dense Optical Flow. It demonstrates the use of the calcOpticalFlowFarneback function. Again it makes use of the previous CVImage object to bridge between the Processing PImage and OpenCV Mat. The example also reduces the size of the video (using the variable factor) before sending it for the optical flow processing; otherwise, the process can be lengthy.
The optical flow process will basically compare two consecutive frames (the Mat last and grey) from the live webcam video. It will try to compute where the current pixels move to in the new frame.
Here are a number of screen shots from the sample run.
The example again adopts the CVImage to demonstrate the use of blur filter. It uses the medianBlur() function from the Imgproc module in OpenCV. Detailed description can also be found in the OpenCV tutorial here.
In this example, I demonstrate the use of the image processing module (Imgproc) with the filter2D function. It is similar to the generic Photoshop filter where you can design the convolution matrix. The example references the OpenCV tutorial and uses a sharpen filter. It also uses the CVImage class in the last post.
Here is a screen shot of the Processing window.
