Fast and Robust Realtime Storefront Logo Recognition

Storefront brand image (logo) recognition has many useful applications. Various methods have been proposed using local and global feature matching; however, logo recognition continues to be a challenging area; unlike nature scene images, which are rich in texture details, brand images often lack texture variation, and therefore provide fewer feature points for matching. In addition, reference and test images may be acquired with varying levels of rotation, resolution, size, quality, and illumination. In this paper, I present a highly efficient and accurate approach for recognizing logos in storefront images. I separate the algorithm into two primary phases representation and recognition, each of which deals with a separate portion of the logo recognition algorithm. After training the logo recognition system with 381 pre-segmented storefront logos, I was able to achieve an 86.0% classification accuracy on 100 test images taken from a wide variety of locations.