彩色图像快速检索方法的改进研究与仿真

Color Image Retrieval Methods Fast Improvement Research and Simulation

对彩色图像进行快速检索,可以减少冗余的图像信息,提高彩色图像处理效率。对颜色空间进行有效分割,实现颜色量化,获取彩色图像颜色特征是彩色图像检索的重要环节。但传统的基于目标区域的算法只能根据图像的灰度分量,提取图像的纹理特征,缺少相关的颜色特征提取参数,降低了彩色图像检索的精度与时效性。现提出一种新的彩色图像快速检索方法,利用MATLAB语言编写基于颜色的彩色图像检索程序实现颜色特征的提取,利用等间距离化方法对HSV颜色空间进行有效分割,实现颜色量化,将颜色特征表现在直方图中。通过基于划分的方法描述颜色的空间分布信息,利用欧几里德距离公式计算待检索图像同数据库中目标图像间的相似性,根据相似性的大小完成检索。仿真结果表明,改进方法可以在大量的图像信息中快速、准确的提取出所需的具有图像特征的彩色图像,识别过程具有实时性和高精度准确性,证明改进算法具有较强的可行性。

In this paper,we proposed a new fast retrieval method for color images. Firstly,we wrote color image retrieval program based on colors to extract color feature using MATLAB language. Then we segmented HSV color space effectively to achieve color quantization using equal interval separation method and presented color feature in histogram. Color spatial distribution information was described through division based method. We also used Euclidean distance formula to calculate similarity between retrieval images and target images. Finally,we completed retrieval according to the similarity. The simulation results show that the modified method can extract color images with required features quickly and accurately in vast image information. The recognition progress is of timeliness and high precision. It proved that the algorithm mentioned above has fine feasibility.