Persian logo recognition using local binary patterns

Abstract

Nowadays, image processing is getting more popular due to the daily increase of diverse data acquisition methods such as digital scanners and cameras. Due to the high volume of archived documents, automatic document classification methods can help to save the time and space in digital document organization. Logos in official and business documents are used to identify document identities. Different approaches have been used for logo recognition yet, many of which has complex computations to achieve a high level of precision. In this paper, a novel algorithm for accurate logo recognition with low level of computational complexity is proposed based on Local Binary Pattern (LBP). We proposed PerLogo dataset consisting 850 images of 10 different classes of logos has been proposed in this paper. Through 3 separate experiments over 50, 60, 70 images per each class the proposed system has been evaluated. Experimental results show that recognition rate is increased with increasing the number of training images per class. Experimental results show the recognition accuracy of 98% when 0.09 salt and pepper noise are added to the test images, which is more than 95% accuracy proposed by the state-of-the-art approaches achieving 95% accuracy.