摘要
在图像检索领域中,形状特征是图像的最重要视觉特征之一,利用形状特征进行检索可提高检索的准确性,但形状边界的自动提取一直是图像处理领域多年的难题。为了提高图像精度和准确性,提出一种基于轮廓检测的图像检索方法,首先用色彩聚类的方法对图像进行预处理,对有意义的聚类区域进行边缘追踪,然后采用基于Snake轮廓检测的算法完成图像分割,提取底层形状特征并用傅立叶描述子加以描述,进行相似度匹配。引入支持向量机的相关反馈算法来提高检索精度。实验结果表明了方法的有效性。
The shape feature of images is one of the most important visual features. It can be used to improve the accuracy of image retrieval. However automatic extraction of the border shape has long been a difficult problem in image processing. A new image retrieval method based on contour detection is presented. First , a method of color clus- tering is employed to pre - process the image, and contour - tracking is used to the meaningful clustering region. Then an arithmetic based on Snake for contour detection is invoked to implement the image segmentation. The low - level feature of the shape is subsequently extracted and described by Fourier descriptor for similar matching. At last, to improve the precision of image retrieval, a relevance feedback mechanism based on support vector machines is invoked. The experiment results show that this method is effective.
出处
《计算机仿真》
CSCD
北大核心
2009年第10期270-273,293,共5页
Computer Simulation
关键词
图像检索
色彩聚类
轮廓检测
傅立叶描述子
支持向量机
Image retrieval
Color clustering
Contour detection
Fourier descriptor
Support vector machines（ SVM ）
