K-means聚类算法在鞋身色彩搭配方面的应用研究

Application Research of K-means Algorithm in Shoe Color Combinations

颜色是消费者选择时尚产品时最重要的因素之一,对鞋类图片中鞋体的颜色搭配的提取和色彩比例的计算是流行趋势分析的重要一环。使用K-means聚类算法对鞋的颜色进行聚类,并详细对比了采样、样本加权、均值排序三种预处理方法进行加速处理后的K-means算法在不同K取值下的性能。最后利用该算法在Lab色彩空间中对鞋身图片进行配色方案聚类提取。结果表明改进后的K-means算法比朴素版K-means算法快9~14倍,且能准确识别鞋身的主要基础色与每种基础色的比例。

Color is one of the most important factors for consumers to choose fashion products.Extracting and calculating the color combinations are essential for fashion trend analysis.In this study,K-means algorithm was used to cluster the color of shoes,and three pretreatment methods were performed to accelerate K-means under different K values,including sampling,sample weighting and mean sorting.The results show that the improved K-means algorithm is 9-14 times faster than the naive version,and can accurately identify the proportion of each main base color of the shoe.