Font Size: a A A

The Visibility Of The License Plate Image Enhancement And Tilt Correction

Posted on:2017-02-26Degree:MasterType:Thesis
Country:ChinaCandidate:C XuFull Text:PDF
GTID:2348330488985152Subject:Transportation engineering
Abstract/Summary:PDF Full Text Request
In recent years, the increasing demand for transportation causes more road traffic conflicts. Intelligent transportation has become an important role in the entire transport ecological environment. Of which the most widely used automatic number plate recognition technology is an important part, in order to provide a more license plate recognition quality image sources, this article will focus on two aspects, to enhance the visibility and tilt correction.First, in terms of enhancing the visibility of the license plate image, this paper aimed to eliminate image noise that bring the visual impact. In the image denoising nonlocal framework, the use of low-rank model nuclear norm minimization and image of a priori information by the weighting coefficient reflected. Article utilizes the larger singular value matrix having a high priority on the role of this nature and to a large singular value to a smaller weight to achieve less shrinkage proposed weighted nuclear norm minimization for image denoising. Experiments demonstrate that the proposed method and subjective visual details are given on the overall advantages compared to the previous method, and peak signal to noise ratio has been higher value.The image tilt correction is an important step in the character segmentation and even the recognition. As a consequence, we proposed a wider range of license plate tilt correction based on transform invariant low-rank textures. According to different environmental conditions, to tilt correction for images which is randomly shooting, experiments show that even in the face of some harsh conditions, such as no border, pollution, ambient occlusion, image noise, the method of article also obtained excellent correction results. This paper put forward a new idea of tilt correction by means of the image overall information, which has abandoned the previous point of view from the corner, edge and other local features, and therefore it has a higher robustness and accuracy. Extensive experimental results demonstrate that the effectiveness of the method in the article and can satisfy the demands of the whole vehicle license plate recognition system in real time.
Keywords/Search Tags:image denoising, nuclear norm minimization, license plate tilt, low-rank texture, affine transformation
PDF Full Text Request
Related items