|
|
Fast Reducing Redundancy in Capsule Endoscopy Video Based on SURF |
Liu Xiaoyan1*, Gong Junhui1, 3, Li Xiangdong2, Wu Weining1, Peng Tongsheng1 |
1(College of Electrical and Information Engineering, Hunan University, Changsha 410082, China) 2(Suzhou OIH Medical Scientific LTD. CO., Suzhou 215000, Jiangsu, China) 3(College of Electrical and Information Engineering, Hunan Institute of Engineering, Xiangtan 411101, Hunan,China) |
|
|
Abstract Capsule endoscopy captures tens of thousands of images through the whole digestive system with a large number of redundant images. To determine abnormal images would consume considerable time and effort of medical doctors by checking each frame. Therefore, it is meaningful to screen out the redundant images automatically. In this paper, feature points and the corresponding feature vector was extracted in successive capsule endoscopy frames with SURF. Vector angular was applied to evaluate the matching degree between two feature points from the successive frames. Maximum between-class variance was used to identify the matched feature points adaptively. Finally, the quantity of matched feature points and block matching degree was used as the index to screen out redundant images. Experimental tests in captured 128 videos by capsule endoscopy demonstrate that the proposed approach has a rapid processing speed of 0.06 seconds/frame, with good performance in the recall, precision and F-measure index (81%, 90%, and 85%, respectively). Moreover, it shows good robustness to uncertainties such as peristalsis, illumination and bubbles.
|
Received: 31 July 2015
|
|
|
|
|
[1] Gong Feng, Swain P, Mills T. Wireless endoscopy [J]. Gastrointestinal Endoscopy, 2000, 51(6): 725-729. [2] Iddan G, Meron G, Glukhovsky A, et al. Wireless capsule endoscopy [J]. Nature, 2000, 405-417. [3] Koulaouzidis A, Iakovidis DK, Karargyris A, et al. Wireless endoscopy in 2020: Will it still be a capsule? [J]. World J Gastroentero, 2015, 21(17): 5119-5130. [4] Lee HG, Choi MK, Shin BS, et al. Reducing redundancy in wireless capsule endoscopy videos [J]. Comput Biol Med, 2013, 43(6): 670-682. [5] 马丽黎,陈世耀,姚礼庆. 穿越肠道的照相机-胶囊内镜[EB/OL]. http://gi.cmt.com.cn/detail/22401.html, 2011-05-16 /2015-07-21. [6] Liu Hong, Pan Ning, Lu Heng, et al. Wireless capsule endoscopy video reduction based on camera motion estimation [J]. J Digit Imaging, 2013, 26(2): 287-301. [7] Iakovidis DK, Tsevas S, Polydorou A. Reduction of capsule endoscopy reading times by unsupervised image mining [J]. Comput Med Imaging Graph, 2010, 34(6): 471-478. [8] Iakovidis DK, Koulaouzidis A. Software for enhanced video capsule endoscopy: challenges for essential progress [J]. Nat Rev Gastro Hepat,2015, 12(3): 172-186. [9] Li Baopu, Meng MQ, Hu Chao. Motion analysis for capsule endoscopy video segmentation [C]//2011 IEEE International Conference on Automation and Logistics. Taipei: IEEE, 2011:46-51. [10] Karargyris A, Bourbakis N. A video-frame based registration using segmentation and graph connectivity for Wireless Capsule Endoscopy [C]//Life Science Systems and Applications Workshop. Maryland: IEEE, 2009:74-79. [11] 潘宁. 胶囊内镜图像序列冗余数据筛查方法研究 [D]. 武汉:华中科技大学, 2013. [12] Liu Ce, Yuen J, Torralba A, et al. SIFT flow: Dense correspondence across different scenes [C]//Proceedings of the 10th European Conference on Computer Vision: Part III. Marseille: Springer, 2008:28-42. [13] Bay H, Tuytelaars T, Gool LV. SURF: Speeded up robust features [M]//Computer Vision-ECCV 2006. Berlin: Springer-Verlag, 2006:404-417. [14] 李秀艳, 刘铁根, 邓仕超, 等. 基于SURF算子的快速手背静脉识别 [J]. 仪器仪表学报, 2011, 32(4): 831-836. [15] 吴一全, 沈毅, 陶飞翔. 基于NSCT 和 SURF 的遥感图像匹配 [J]. 遥感学报, 2014, 18(3): 618-629. [16] Lowe DG. Distinctive image features from scale-invariant keypoints [J]. Int J Comput Vision, 2004, 60(2): 91-110. [17] Lowe DG. Object recognition from local scale-invariant features [C]//Proceedings of the International Conference on Computer Vision (ICCV 1999). Corfu: IEEE, 1999:1150-1157. [18] Mikolajczyk K, Schmid C. Indexing based on scale invariant interest points [C]// International Conference on Computer Vision (ICCV 2001). Montbonnot: IEEE, 2001:525-531. [19] Lindeberg T. Feature detection with automatic scale selection [J]. Int J Comput Vision, 1998, 30(2): 79-116. [20] Neubeck A, Van GL. Efficient non-maximum suppression [C]//Proceedings of the 18th International Conference on Pattern Recognition. Hong Kong: IEEE, 2006:850-855. [21] Otsu N. A threshold selection method from gray-level histograms [J]. IEEE Trans Syst Man Cybern B, 1979, 9(1): 62-66. [22] Lee HG, Choi MK, Lee SC. Motion analysis for duplicate frame removal in wireless capsule endoscope [C] //Society of Photo-Optical Instrumentation Engineers Conference Series. Florida: SPIE, 2011: 63-66. [23] Nancy C. MUC-4 evaluation metrics [C]//The Fourth Message Understanding Conference. Virginia:MCL, 1992: 22-29. |
|
|
|