主颜色提取算法及其在服装图像检索中的应用

本文提出一种主颜色提取算法用于服装图像的颜色特征提取。该算法的基本思想是:首先根据服装的主颜色占比情况判断该服装是花色还是非花色，将花色服装用特征值进行标记。接着，将颜色直方图中的特征值按占比大小降序排列，去除占比较小的部分特征值，保留主颜色特征值。在进行服装检索时，先对数据库中的图像进行第一轮筛选，将花色与非花色服装进行分类；然后再对分类后的服装根据主颜色特征值进行第二轮筛选，得到最后的检索结果。实验结果表明，无论对于简单背景还是复杂背景的服装，无论是花色还是非花色服装，本文提出的算法相比于现有的其他算法，服装检索的查全率和查准率都得到了显著提高。

Main Color Extraction Algorithm and its Application to Clothing Image Retrieval

The main color extraction algorithm is proposed to extract the color feature of clothing image in this paper. The idea of this algorithm is to firstly judge the clothing is whether flowery or non flowery according to the main color of the garment, and then arrange the eigenvalues of color histogram based on descending order ,remove the smaller part of the eigenvalues and keep the main color eigenvalues. In clothing image retrieval, firstly, we should make the screening of the image in the database according to the clothing image is whether flowery or non flowery ; and then the second round selection was conducted according to the main color characteristic value, and the final retrieval results were obtained. The experimental results show that the recall ratio and precision ratio of the the improved algorithm compared with other existing color feature extraction methods, whether the clothing of simple background or complex background, flowery or non flowery, has obvious advantages.