Clothing image feature extraction based on SURF

2018 International Conference on Advanced Electronic Materials, Computers and Materials Engineering

材料科学;无线电电子学;计算机科学

Chen, L.L.^1 ; Han, R.P.^2 ; Meng, Q.C.^1

School of Information Engineering, Beijing Institute of Fashion Technology, Beijing

100029, China^1

Chinese Fashion and Technology Research Institute, Beijing Institute of Fashion Technology, Beijing

100029, China^2

关键词: Feature extraction algorithms;    Image feature extractions;    Input sample;    Machine learning methods;    Recognition accuracy;    Speeded up robust features;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/439/3/032104/pdf
DOI  :  10.1088/1757-899X/439/3/032104

来源: IOP

【 摘 要 】

The SURF (Speeded Up Robust Features) is one of the most commonly used artificial feature extraction algorithms and has a good robustness. SURF is widely used in image processing and machine vision. This paper introduces the implementation of SURF in a more comprehensive way, and applies it to the feature extraction of clothing images. The clothing features extracted by using SURF can be applied in clothing classification, identification, retrieval and matching in combination with machine learning method. It can also be used as input sample of deep neural network to improve classification or recognition accuracy.