An improved vehicle-logo localization algorithm based on texture analysis

Abstract

Vehicle-logo location is the key link in vehicle-logo recognition, and it directly impacts the effect of vehicle-logo recognition. However, because of its inherent particularity, vehicle-logo location is a very challenging. In this paper, we proposed an improved vehicle-logo location algorithm based on texture analysis. Firstly, we locate the vehicle-logo coarsely according to the prior knowledge of the position relationship between the vehicle-logo and the license plate; Secondly, we use the texture measure to detect the texture type of the scattered network; Thirdly, we use the Laws texture mask to suppress the scattered network texture. The concrete practice is to carry on the vertical edge template to filter out the horizontal texture, and carry on the transposed matrix of the vertical edge template to filter out the vertical texture; Finally, we can locate the vehicle-logo accurately by mathematical morphology processing. In this paper, the filtering method using the Laws texture template remains more texture information, and improves the algorithm robustness for the night, and at the same time, it has a good location effect for the dark vehicle-logo.