摘要
物体形状的特征提取是图像检索与识别中的重要研究内容。考虑到物体形状的多变性,给出了一种基于多模复杂网络模型的形状特征提取方法。首先,以形状边界轮廓点作为节点,利用节点之间的欧氏和内部距离作为连接顶点之间边的权值构建初始网络模型;然后,分别基于欧氏距离和内部距离对初始网络进行演化;最后,从子网络中提取特征,并进一步用于形状描述。检索与分类实验表明,所提方法相比于传统的单一模态下的复杂网络特征提取方法具有更强的鲁棒性。
Shape feature extraction is important in content based image recognition. Considering the variance of shape, this paper presents a new shape feature extraction method based on multi-model complex network. Firstly, an initial complex network is constructed with nodes corresponding to the boundary points and edges allocated with Euclidean and inner distances as weights existing between each node pairs. Then, this initial network is evolved based on Euclidean and inner distances, respectively. At last, some features are extracted from these sub-networks and further provided for the shape. Experimental results on both object classification and retrieval show that the proposed method is more robust compared with the single-model complex network model.
出处
《计算机科学与探索》
CSCD
2013年第6期570-576,共7页
Journal of Frontiers of Computer Science and Technology
基金
国家自然科学基金Nos.61202228
61073116~~
关键词
特征提取
复杂网络
内部距离
动态演化
feature extraction
complex network
inner distance
dynamic evolution
