A license plate location algorithm based on multi-component edge combination of the HSI color space

Abstract

Proposed a novel algorithm based on combination of multi-component edge license plate location in the HSI color space. First, the RGB image is converted to HSI image, and next we do edge detection on the S component and I component of the HIS color space to get the edge image of two different images, then do AND operation on the two images, and use the H component of the HIS color space to determine the threshold of edge image. After that, we use the method of graphics morphology to process the edge image, and according to aspect ratio and area features of license plate image we can finally accurately locate the plate region. We did experiments on two hundreds different light conditions and background of vehicle image. Experimental results show that, the location accuracy of this method is up to 96.5% and the period of location is less than 250ms.