【作者】 李太君； 康耀红； 温小斌； 潘学松；

【机构】 海南大学信息科学技术学院；

【摘要】 <正>1前言颜色是图像最显著的视觉性质,也是人们识别图像的主要感知特征,文[1]提出了利用全局颜色直方图进行图像检索的基本思想。全局颜色直方图描述的是各种色彩在整幅图像中出现的频度及所占的更多还原

【Abstract】 Description of the global color histogram is the proportion of various of colors in the whole image.It can not reflect the spacial distributing information of colors.In order to remedy this defect,a new image retrieval algorithm: sector model is presented in this paper.By ameliorating the HSV color model,the hue,saturation and value (intensity) is projected respectively onto the radial and angle directions,and six projective histograms are obtained. The projective histograms of hue and saturation characterize the color distribution in space,and the projective histogram of value reflects the shape of the image in a certain extent.Then,take the first three central moments(mean, variance,skewness) of these six projective histograms as features to compute the distances between images.Experimental results show that this algorithm performs tersely and quickly and has better recall rate and precision.更多还原