A real-time face to camera distance measurement algorithm using object classification

In a human and computer interaction based system, distance estimation by computer vision between camera and human face is a vital operation. To calculate the distance between the camera and face, an estimation method based on feature detection is proposed in this paper, where detection of eyes, face and iris in an image sequence is described. From the estimated iris and the distance between the centroid of the iris, an algorithm is proposed to determine the distance from camera to face. An architecture for face detection based system on AdaBoost algorithm using Haar features and Canny and Hugo Transform for edge and circular iris estimation is presented here. Wrongly detected faces are removed by analyzing the disparity map. From the estimated face, canny and Hugo transform is used to determine the iris, and to calculate the distance between the centroid of iris. Later Pythagoras and similarity of triangles are used for distance estimation. The implementation is done in C++ using Intel OpenCV image processing libraries to reduce system overhead.