Research Of Content-based Fashion Image Retrieval

Due to the steady growth of computer, multimedia,and Internet techniques,a huge amount of images are available. Currently, rapid and effective searching for desired images from large-scale image databases becomes an important and challenging research topic.Content-based image retrieval (CBIR) is the set of techniques to address the problem of retrieving relevant images from an image database based on automatically derived image features. In recent years,CBIR is a very active research direction and has been applied to many fields.Now the retrieval system about fashion image rarely can be found at home and abroad. China is a big fashion production country, and it has the power of consumption very crayzed. So the needs of fashion image retrieval system is essential.In this paper, the exploratory research work has been done around the low-level feature extraction, which include shape,texture,color and so on. The main contributions of this paper are summarized as follows:(1) We analysed and discussed the low-level feature descriptions including shape texture and color, the similarity measure between the features and the evaluation methods of image retrieval algorithms.(2) An image retrieval algorithm based on the axis of least inertia. It that is capable of preserving both contour as well as region information extracts shape feature points, and uses Weighted Euclidean distance for the terms of multi-valued type of feature points to compute the degree of similarity between two shapes. This method is invariant to image transformations (translation, rotation and scaling), and the shape matching is more accurate.(3) An image retrieval algorithm based on the main direction of texture. Through FFT we got the histogram of the main direction of texture. Then we analysed and researched GLCM in the fashion image retrieval. texture features from the combination of its four arguments as the characteristic retrieval images.Through testing, we can see that these methods for content-based image retrieval have very good results.