• Journal of Korea Multimedia Society
• /
• v.19 no.9
• /
• pp.1647-1658
• /
• 2016

RGB+D database has been widely used in object recognition, object tracking, robot control, to name a few. While rapid advance of active depth sensing technologies allows for the widespread of indoor RGB+D databases, there are only few outdoor RGB+D databases largely due to an inherent limitation of active depth cameras. In this paper, we propose a novel method used to build outdoor RGB+D databases. Instead of using active depth cameras such as Kinect or LIDAR, we acquire a pair of stereo image using high-resolution stereo camera and then obtain a depth map by applying stereo matching algorithm. To deal with estimation errors that inevitably exist in the depth map obtained from stereo matching methods, we develop an approach that estimates confidence of depth maps based on unsupervised learning. Unlike existing confidence estimation approaches, we explicitly consider a spatial correlation that may exist in the confidence map. Specifically, we focus on refining confidence feature with the assumption that the confidence feature and resultant confidence map are smoothly-varying in spatial domain and are highly correlated to each other. Experimental result shows that the proposed method outperforms existing confidence measure based approaches in various benchmark dataset.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.53 no.5
• /
• pp.132-140
• /
• 2016

In this paper, we propose boundary-preserving stereo matching method based on adaptive disparity adjustment using confidence region detection. To find the initial disparity map, we compute data cost using the color space (CIE Lab) combined with the gradient space and apply double cost aggregation. We perform left/right consistency checking to sort out the mismatched region. This consistency check typically fails for occluded and mismatched pixels. We mark a pixel in the left disparity map as "inconsistent", if the disparity value of its counterpart pixel differs by a value larger than one pixel. In order to distinguish errors caused by the disparity discontinuity, we first detect the confidence map using the Mean-shift segmentation in the initial disparity map. Using this confidence map, we then adjust the disparity map to reduce the errors in initial disparity map. Experimental results demonstrate that the proposed method produces higher quality disparity maps by successfully preserving disparity discontinuities compared to existing methods.