基于主色调匹配的图像检索方法

An Image Retrieval Method Based on Dominant Color Matching

在基于内容的图像检索中,颜色作为图像的一种重要视觉信息已得到广泛应用。主色调可以反映一幅图像的基本面貌,其实质就是用较少的颜色来表示原图像,并且在视觉上与原图像相似。由此,利用人眼对颜色的感知进行相似色区间的划分,从而实现对颜色的量化,并利用量化后得到的图像直方图进行图像的相似度匹配,取得了满意的实验效果。

As an important visual info., color is now widely used in image retrieval. The dominant color can reflect the main situation of an image. Its essential is to present the image by means of less color and make little difference between the new image and the original image. Based on humans perception to color, this paper proposes a new dividing method for similar colors in HSI space, which is to quantize color. After that , we can get histogram to search similar images from the image repository. The result of the experiment is satisfying.