A fast color barcode detection method through cross identification on mobile platforms

Abstract

The use of color barcodes enables rich formation can be stored in a 2D scene image. However, automatic detection of color barcodes on a mobile platform is challenging due to the sensitivities to color, light condition, curved product surface and the distortions during camera capturing. In this paper, we present a novel approach for fast and robust color barcode detection in natural scenes. To be generic, we consider a color barcode to be a grid of colored squares consisting of at most 4 or 8 colors. A feature detector for finding crosses formed by differently colored cells is proposed. Next, candidate barcode regions are generated using a thresholding method and those regions containing more than a specific number of crosses are declared to be barcode regions. Comparisons with the base line Hough transform method for matrix code detection are performed, and the experimental results show the efficiency and robustness of the proposed method in the presences of blurry and complex background of scene images.