Abstract | ||
---|---|---|
This paper presents a new algorithm for image based obstacle detection from Unmanned Surface Vehicles (USV) by exploring the global sparseness of image patches. The horizon line is taken as the prior knowledge so as to process the algorithm within the sea area. Considering the geometric relation between the camera and the sea surface, the resolution of the observation that close to the horizon is smaller than that close to the image bottom, thus we adopt the variable size image windows to sample image patches for analysing their global spareness across the whole sampled patches set, in which the less spareness patches are clustered as sea surface, while the ones with high spareness scores are considered as candidates for obstacles. Finally, the candidates far from the mean of the sea surface patches are selected as obstacles. Experiments on our own dataset from the USV demonstrate the efficiency of the proposed algorithm, which shows higher accuracy than the traditional method as well as the state-of-the-art saliency detection method. |
Year | DOI | Venue |
---|---|---|
2015 | 10.1109/IVCNZ.2015.7761552 | 2015 International Conference on Image and Vision Computing New Zealand (IVCNZ) |
Keywords | Field | DocType |
global sparsity potential,unmanned surface vehicles,image based obstacle detection,USV,global image patch sparseness,horizon line,geometric relation,camera,image bottom,variable size image window,image patch sampling,spareness score,sea surface patch,saliency detection method | Obstacle,Computer vision,Pattern recognition,Salience (neuroscience),Computer science,Horizon,Image based,Feature extraction,Artificial intelligence,Image resolution | Conference |
ISSN | ISBN | Citations |
2151-2191 | 978-1-5090-0358-7 | 0 |
PageRank | References | Authors |
0.34 | 5 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Xiaozheng Mou | 1 | 2 | 2.11 |
Han Wang | 2 | 148 | 22.31 |