¹Department of computer Science, National Chengchi University Taipei, Taiwan (R.O.C.)
²Center for Computational Research and Applications, National Chengchi University Taipei, Taiwan (R.O.C.)

---

Abstract—Logos belonging to the same brand can be diverse in shape and color, so logo recognition is a challenging research topic. Because of the diversity of logos, if we consider only the shapes of logos in the recognition process, we will suffer from the high cost in terms of the overall execution time. In this paper, we propose to use shapes as well as colors of logos in the recognition process. Before using shapes of logos to recognize the brand of an interested logo image, we calculate the color difference between it and the brand logo images in the database. Then, we select from the database the images to which the interested logo image is similar according to their color difference, and we measure the similarity between the shape of the interested logo image and the shape of each of the selected images. We recognize the interested logo image as the brand logo that is most similar to it. The experimental results show the following: Compared to that we use only shapes in the recognition process, if we use color difference before using shape, then we can save a significant amount of time, and the proposed approach has the potential to improve the overall recognition rate.

Index Terms—Color difference, logo recognition, shape context

[PDF]

Cite: Yi-Peng Lin, Kuo-Wei Hsu, "Using Color Difference with Shape Context for Logo Recognition," Journal of Software vol. 9, no. 8, pp. 2188-2193, 2014.