We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
Track a random object to move the mouse or control a game
Object Tracking based on color using OpenCV Author: Adarsha Joisa If you have downloaded the source code, please send me an email at adarshajoisa[at]gmail[dot]com. It'd help me keep track of the number of people using this tool. :) I hope you enjoy it and improve it. This is a simple tool to track random colored objects using the computer's webcam. I've developed it on my laptop, so the default webcam has been used. If you are using a different(external) webcam,change the line: "capture = cvCaptureFromCAM(0);" by replacing the "0" with the appropirate index for your webcam. The algorithm for color detection is as follows: 1. A capture is created to capture video from the webcam. A yellow square (20x20 pixels) is displayed at the center of the screen. This is considered to be the ROI. 2. The user shows the object to the camera such that the object fills the yellow square, and hits Space. 3. The current frame is captured. We extract the most prominent color in the ROI using the getHSVranges() function. 4. Once we get the HSV ranges, the camera starts capturing live video. The frames are thresholded using the range we got earlier. (NOTE: I've written my own function for thresholding, but the openCV function cvInrangeS() works in the same way.) 5. By calculating the moments for the thresholded image, we get the center of the object. (Note: If there's anything else in the camera's field of view with a color similar to the object, this center will be offset towards the colored region. Hence, it's best to use some object with a bright and rare color. (I've tried objects of red, orange, yellow, blue and purple, and they worked well.)) The position of the center is used to perform any useful function. In this program, I've provided 2 options, one to control the mouse pointer and second to simulate the arrow keys of the keyboard. The arrow keys are simulated by just moving the object in similar directions: left for left, up for up, left and up for left+up, and so on. The input is sent to the currently active window. You can comment out these lines to hide the webcam video and thresholded video windows: cvShowImage("thresh", imgThresh); cvShowImage("video", frame); Note: compile on linux as: g++ -ggdb `pkg-config --cflags opencv` -o track TrackColour.cpp `pkg-config --libs opencv` Note: You need to have OpenCV development libraries installed on your system to compile this code. Refer to http://http://opencv.willowgarage.com/wiki/ for instructions on installing opencv.