基于颜色和纹理特征的图像检索技术
首发时间:2009-08-05
摘要:提出了一种基于颜色和纹理的图像检索方法:利用等面积圆环划分的方法对图像进行分割,再提取各个圆环的颜色直方图,这种方法即可描述颜色的空间分布信息,还能较好的保持颜色的精度,然后将分块主颜色和分块颜色矩组合起来实现颜色检索;在纹理检索方面,利用连续小波变换对图像进行分解,求出分解后每个子带小波系数的均值和标准差,通过特征向量提取图像的纹理特征;相似性度量采用曼哈顿距离。最后综合运用颜色和纹理特征对图像进行检索,实验结果表明,提高了检索的查全率和查准率。
For information in English, please click here
Image Retrieval Based on Color and Texture Feathers
Abstract:An image retrieval method based on color and texture feture is proposed in the paper. Dividing method of equal area cirque is used to segment image, and then extract the color histogram of each ring. This method can describe the distributing information of color space, but also maintain color accurary well. The main color of sub-block and color moments are combined to achieve color retrieval; in texture retrieval, with the use of continuous wavelet transform, image can be decomposed. Work out the mean and standard deviation of each sub-band wavelet coefficients, then texture fetures can be extracted by feature vector. Similarity measure using Manhattan distance. Finally, experimental results show, the integrated use of color and texture features improve the retrieval recall and precision.
Keywords: main color color moments wavelet transforms texture feature
基金:
论文图表:
引用
No.3427848636012494****
同行评议
共计0人参与
勘误表
基于颜色和纹理特征的图像检索技术
评论
全部评论0/1000