Hierarchical Contours Based on Depth Images

In this paper we propose an algorithm for contour extraction that combines information from the two types of images captured with a Kinect camera. One image contains information about the depth and the other contains the color information. The algorithm segments the different levels of depth using the distance image and for each zone it extracts the contours on the color image. We used histogram segmentation. The histogram was smoothed using a Lagrange interpolation. As a final result the algorithm outputs a hierarchical contour that depends on the depth and the color image as well.