Time filter

Source Type

Liu T.,Jiangsu Provincial Key Laboratory of Image Processing and Image Communication | Liu T.,Nanjing University of Posts and Telecommunications | Dai X.,Jiangsu Provincial Key Laboratory of Image Processing and Image Communication | Dai X.,Nanjing University of Posts and Telecommunications | And 6 more authors.
Yi Qi Yi Biao Xue Bao/Chinese Journal of Scientific Instrument | Year: 2012

A fast stereo matching method based on illumination invariant features and modified support weights is proposed to determine the correspondences in different images of the same scene taken from different viewpoints. Firstly, illumination invariant features are constructed densely for the stereo pairs and initial matching costs are calculated from the invariant features; robust matching costs are obtained based on two-pass aggregation with tri-cube weight kernels; and initial disparities are selected using Winner-Takes-All optimization method. Secondly, intelligent disparity inpainting procedures are in turn implemented or executed with modified bilateral filtering based on multi-directional neighbor tri-cube weight kernels, occlusion detection and disparity error concealment. Experiments indicate that this technique with low computational burden, can produce comparably satisfactory dense disparity map, and can improve the adaptability to illumination variations.


Liu T.-L.,Jiangsu Provincial Key Laboratory of Image Processing and Image Communication | Liu T.-L.,Nanjing University of Posts and Telecommunications | Dai X.-B.,Nanjing University of Posts and Telecommunications | Zhu X.-C.,Jiangsu Provincial Key Laboratory of Image Processing and Image Communication | And 2 more authors.
Dianzi Yu Xinxi Xuebao/Journal of Electronics and Information Technology | Year: 2012

This paper presents a hierarchical cost aggregation-based fast stereo image matching method based on Weber's law and guided filtering. Weber local descriptors for each color channel are firstly extracted from stereo pairs, and raw matching costs between the images are initialized by the descriptors. The matching costs are enhanced with guided filtering to extract the subsets of disparity candidates. Joint spatial discrete sampling and adaptive support weight are utilized to implement hierarchical cost aggregation on the candidate subsets. Then initial disparities from the subsets are selected fast and optimally. Modified bilateral filtering and symmetric warping-based post-processing are sequentially exploited in disparity refining to improve effectively ambiguous regions of initial disparity maps. The experimental results indicate that this proposed technique can obtain piecewise smooth, accurate and dense disparity map while eliminating effectively matching ambiguity. Being concise, fast and high efficiency, and it is robust to illumination change.


Liu T.,Jiangsu Provincial Key Laboratory of Image Processing and Image Communication | Liu T.,Nanjing University of Posts and Telecommunications | Liu T.,Nanjing Southeast University | Dai X.,Nanjing University of Posts and Telecommunications | And 8 more authors.
Dongnan Daxue Xuebao (Ziran Kexue Ban)/Journal of Southeast University (Natural Science Edition) | Year: 2012

An integral images-based fast keypoint extraction and matching method being robust to illumination changes is presented. First, robust illumination local features are fast extracted by multi-scale up-sampling and filtering the Riemann integral of the stretched contrast responses using integral images, and multi-scale interest feature keypoints are detected based on local maximum of them in a multi-resolution framework. Secondly, covariance descriptors are constructed with integral images for the center-symmetrical rectangular regions around the obtained points, then the Log-Euclidean metrics are adopted to measure the difference between the covariance matrices around the keypoints from two matching scenes. Finally, automatic matching between two feature point sets is implemented with the sparse matching strategy. Experimental results prove that the proposed method is of simple structure, fast and efficient computation with high accuracy, while being adaptive for illumination changes.

Loading Jiangsu Provincial Key Laboratory of Image Processing and Image Communication collaborators
Loading Jiangsu Provincial Key Laboratory of Image Processing and Image Communication collaborators