Search for: [Abstract = "In this paper, two popular methods of scene disparity estimation are compared for application in augmented reality. Augmenting computer rendered objects between real objects is challenging problem. Especially when image comes from video sequence or real time video. The proposed approach is base on the methods of disparity estimation for all pixels using stereo cameras. The group of pixels representing one object usually can be aggregated to one planar layer. Given the depth of any layer, the virtual object can be blended between the layers. Experimental results shows that using stereo calibrated cameras for scene depth estimation is effective. Graph cut method gives more accurate results then block matching method at a cost of lower speed."]

Number of results: 1