Title: Traffic signs identification by using sparse image decomposition and adaptive correlation techniques
Authors: Qinfu Qiu; Qi Zhang; Xiong Chen
Addresses: Electric Engineering Department, School of Information Science and Technology, Fudan University, Shanghai, China ' Electric Engineering Department, School of Information Science and Technology, Fudan University, Shanghai, China ' Electric Engineering Department, School of Information Science and Technology, Fudan University, Shanghai, China
Abstract: Image recognition is an important technology in the present. This paper proposes a new method of image recognising. This approach includes using the morphological component analysis of texture and cartoon layers in image preprocessing, where the matrix of dictionary which is used in the algorithm comes from independent component analysis. Then, use of the adaptive filtering algorithm, which is based on the matching the feature of the standard image and the feature of the original image, highlights the content of the image which is needed to be focused on. In this paper, we select the car camera caught pictures and display the identification effects of the stop sign as an example. Quantitative analysis of the experimental results shows that this proposed method has quite good results and accuracy. At the end of the paper, several recommendations point out its feasible development direction in the future.
Keywords: traffic signs identification; sparse image decomposition; grey; adaptive correlation; image recognition; image preprocessing; independent component analysis; ICA; adaptive filtering; traffic lights; stop signs; morphological component analysis; MCA; cartoon layers; texture layers; image greying.
International Journal of Wireless and Mobile Computing, 2014 Vol.7 No.1, pp.78 - 83
Received: 15 Jun 2013
Accepted: 03 Jul 2013
Published online: 27 Jan 2014 *