关键词: 局部, 几何不变矩, 草图, 检索, 蚁群算法

Abstract: One challenge in sketch-based image retrieval is the effective recognition of different scales, position, rotation and deformation images. To more accurately identify images of different scales, positions and rotations, a sketch-based image retrieval approach using local moment invariant is proposed. First, the geometric characteristics of images are used to determine the coordinate system of image. Then geometry moment invariant is calculated as eigenvector of image blocks which are divided from over all image based on the coordinate system. Following this, the similarity between query sketch and images in database is calculated based on Euclidean distance. Finally, retrieval results are obtained from the similarity ranking and optimized according to ant colony optimization. Experimental results show that the proposed method not only can recognize images after translation, scaling and flipping transformation, but also endure a certain degree of deformation.

Key words: local, geometry moment invariant, sketch, retrieval, ant colony algorithm