In content-based clothing image retrieval, color features can best reflect the basic characteristics of clothing, and also the most stable visual features. Compared with other image features, color features have smaller size, orientation and visual dependence. This paper studies the application of dominant color extraction algorithm in clothing image retrieval, and proposes a clothing classification method based on dominant color ratio. Clothing image is divided into color clothing and non color clothing. On this basis, a main color extraction algorithm of clothing image color feature extraction is proposed. Taking the clothing color features as an example, the image features are analyzed, and then the SVM image classification algorithm is designed to analyze the image features. Then an improved scheme based on data mining technology is proposed, and the analysis model based on association rules is established. Finally, a method of standard man hour correction based on association rules is proposed. The experimental results show that, compared with the existing algorithms, the recall rate and accuracy rate are significantly improved for the clothing with simple or complex background, pattern and non pattern clothing. Analyze and divide the specific areas of clothing image, extract the main color of clothing image, share and recommend clothing image and color extraction results. This research not only has certain research significance, but also has certain practical application value.
Designing a decompositional rule extraction algorithm for neural networks with bound decomposition tree