Vehicle Logo Recognition Using Proposed Illumination Compensation and Six Local Moments

A logo is a significant sign of a vehicle, and it can be considered one of its identification characteristics which cannot be easily changed. Logo identification has many applications, especially in intelligent transportation systems. This paper presents a new method to identify the manufacturer of a vehicle despite challenging images of logos that differ in size, rotation angle and brightness conditions. Three major steps are made to accomplish the recognition task. Preprocessing includes converting images to a grayscale level, then the proposed local illumination compensation is applied. For the feature extraction stage, six local spatial moments are extracted from the sub images of the logo. Finally, a classification process is done using distance measurements. Different tests are made to measure the accuracy rates of the proposed recognition system and to determine the effect of every single feature on the recognition rate, as well as to understand the effect of the proposed local illumination compensation. A dataset containing 544 images for 34 different vehicle logo classes was used, in which each class contains 16 samples. The results show that the success rate reached 86.94853% when all test samples were used in a comprehensive test.