基于感知大范围人造对象图像结构特征抽取.pdfVIP

35 1 () Vol. 35 No. 1 2007 2 Journal of Fuzhou Univers ity( Natural Science) Fe . 2007 : 1000- 2243( 2007) 01- 0035- 06 , ( , 350002) : ( CB IR), . , . (: ), ( perceptual group ing), , . : ; ; ; ; : TP391. 41 : A Structure feature extraction of images containing largemanmade objects based on perceptual grouping YANG Peng, YE Shao- zhen ( College ofM athem atics and Compu ter Science, Fuzhou University, Fuzhou, Fujian 350002, Ch ina) Abstract: Content- ased mi age retrieval ( CBIR) system s have traditionally used color feature and texture feature for low - level mi age analyses, w hich cannotmeetwholly w ith people. s perform ance need. To mi prove the need for ro ust of CBIR, it is very necessary to analyzemore low - level mi - age features in the mi age retrieval system. The paper applies perceptual group ing rules to extract structure feature of mi ages containing large manmade o jects such as uildings, tow ers, ridges, and other arch itecturalo jects. A t last, structural feature is employed to classify special mi age data- ase and valued a little. Keywords: CBIR; perceptual grouping; manmade o ject; structure feature; classification , , , . Iq al andAggarw al [ 1, 2] , , LU / 0. Zhou, [ 3] Ru i and HuangW F(W ater- F illing) . W F , . L i and [ 4] Shap iro ( consistent line clusters) . , , . . , , (: ). , . : 2006- 04- 24 : ( 1980- ), , ; : , , . # 3