Dataset Information

Cross-trees, Edge and Superpixel Priors-based Cost aggregation for Stereo matching.

ABSTRACT: In this paper, we propose a novel cross-trees structure to perform the nonlocal cost aggregation strategy, and the cross-trees structure consists of a horizontal-tree and a vertical-tree. Compared to other spanning trees, the significant superiorities of the cross-trees are that the trees' constructions are efficient and the trees are exactly unique since the constructions are independent on any local or global property of the image itself. Additionally, two different priors: edge prior and superpixel prior, are proposed to tackle the false cost aggregations which cross the depth boundaries. Hence, our method contains two different algorithms in terms of cross-trees+prior. By traversing the two crossed trees successively, a fast non-local cost aggregation algorithm is performed twice to compute the aggregated cost volume. Performance evaluation on the 27 Middlebury data sets shows that both our algorithms outperform the other two tree-based non-local methods, namely minimum spanning tree (MST) and segment-tree (ST).

SUBMITTER: Cheng F

PROVIDER: S-EPMC4448781 | biostudies-literature | 2015 Jul

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Cross-trees, Edge and Superpixel Priors-based Cost aggregation for Stereo matching.

Cheng Feiyang F Zhang Hong H Sun Mingui M Yuan Ding D

Pattern recognition 20150701 7

In this paper, we propose a novel cross-trees structure to perform the nonlocal cost aggregation strategy, and the cross-trees structure consists of a horizontal-tree and a vertical-tree. Compared to other spanning trees, the significant superiorities of the cross-trees are that the trees' constructions are efficient and the trees are exactly unique since the constructions are independent on any local or global property of the image itself. Additionally, two di ...[more]

PMID: 26034314

Similar Datasets

Project description:The fact that natural beta-sheet proteins are usually soluble but that fragments or designs of beta structure usually aggregate suggests that natural beta proteins must somehow be designed to avoid this problem. Regular beta-sheet edges are dangerous, because they are already in the right conformation to interact with any other beta strand they encounter. We surveyed edge strands in a large sample of all-beta proteins to tabulate features that could protect against further beta-sheet interactions. beta-barrels, of course, avoid edges altogether by continuous H-bonding around the barrel cylinder. Parallel beta-helix proteins protect their beta-sheet ends by covering them with loops of other structure. beta-propeller and single-sheet proteins use a combination of beta-bulges, prolines, strategically placed charges, very short edge strands, and loop coverage. beta-sandwich proteins favor placing an inward-pointing charged side chain on one of the edge strands where it would be buried by dimerization; they also use bulges, prolines, and other mechanisms. One recent beta-hairpin design has a constrained twist too great for accommodation into a larger beta-sheet, whereas some beta-sheet edges are protected by the bend and reverse twist produced by an Lbeta glycine. All free edge strands were seen to be protected, usually by several redundant mechanisms. In contrast, edge strands that natively form beta H-bonded dimers or rings have long, regular stretches without such protection. These results are relevant to understanding how proteins may assemble into beta-sheet amyloid fibers, and they are especially applicable to the de novo design of beta structure. Many edge-protection strategies used by natural proteins are beyond our current abilities to constrain by design, but one possibility stands out as especially useful: a single charged side chain near the middle of what would ordinarily be the hydrophobic side of the edge beta strand. This minimal negative-design strategy changes only one residue, requires no backbone distortion, and is easy to design. The accompanying paper [Wang, W. & Hecht, M. H. (2002) Proc. Natl. Acad. Sci. USA 99, 2760-2765] makes use of the inward-pointing charge strategy with great success, turning highly aggregated beta-sandwich designs into soluble monomers.

Project description:Segmentation of liver in abdominal computed tomography (CT) is an important step for radiation therapy planning of hepatocellular carcinoma. Practically, a fully automatic segmentation of liver remains challenging because of low soft tissue contrast between liver and its surrounding organs, and its highly deformable shape. The purpose of this work is to develop a novel superpixel-based and boundary sensitive convolutional neural network (SBBS-CNN) pipeline for automated liver segmentation. The entire CT images were first partitioned into superpixel regions, where nearby pixels with similar CT number were aggregated. Secondly, we converted the conventional binary segmentation into a multinomial classification by labeling the superpixels into three classes: interior liver, liver boundary, and non-liver background. By doing this, the boundary region of the liver was explicitly identified and highlighted for the subsequent classification. Thirdly, we computed an entropy-based saliency map for each CT volume, and leveraged this map to guide the sampling of image patches over the superpixels. In this way, more patches were extracted from informative regions (e.g. the liver boundary with irregular changes) and fewer patches were extracted from homogeneous regions. Finally, deep CNN pipeline was built and trained to predict the probability map of the liver boundary. We tested the proposed algorithm in a cohort of 100 patients. With 10-fold cross validation, the SBBS-CNN achieved mean Dice similarity coefficients of 97.31??±??0.36% and average symmetric surface distance of 1.77??±??0.49?mm. Moreover, it showed superior performance in comparison with state-of-art methods, including U-Net, pixel-based CNN, active contour, level-sets and graph-cut algorithms. SBBS-CNN provides an accurate and effective tool for automated liver segmentation. It is also envisioned that the proposed framework is directly applicable in other medical image segmentation scenarios.

Dataset Information

Cross-trees, Edge and Superpixel Priors-based Cost aggregation for Stereo matching.

Publications

Cross-trees, Edge and Superpixel Priors-based Cost aggregation for Stereo matching.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets