Research Of Uneven Illumination Document Image Correction Method Based On Retinex

With the further popularization of portable high resolution devices,document image has gradually become an important information media to transmit information,and most of document images need to be recognized and converted into a common document file to be better used and stored.In many word recognition technology,OCR(Optical Character Recognition,optical character recognition)technology is the most widely used and the most efficient way to extract text information.However,in real life,we could shoot and collect document image in different light conditions,resulting in the image quality is uneven in quality.Furthermore,it has a badly influence on the recognition rate of OCR.In all kinds of lighting conditions,the influence of uneven illumination is the most prominent and the most likely to occurAiming at the problem of uneven illumination,this paper proposes a new method based on Retinex algorithm.The core idea is to combine the image enhancement and binarization algorithm to process the collected image,in order to ensure the algorithm processing time as quickly as possible under the condition of obtaining clear binarization image,so that the final recognition rate and time performance can reach the balance.At first,this paper summarizes and introduces the current situation of the research on the uneven image at home and abroad,and analyzes the advantages and disadvantages of each method.Then,the key image processing technology which is needed for the document image is described briefly,including image enhancement,image binarization and image denoising.In the third part,this paper present the various illumination conditions that may appear in the document image,and analyze the image information.On the basis of this,the design idea,the evaluation and the validity of the algorithm are described in detail.The fourth part describes the whole process of the algorithm and the details of each part.In the latter part,the effectiveness of the algorithm is verified by testing and analyzing the experimental results.The main reference conditions for the analysis of the experimental results are the OCR recognition rate and algorithm correction time.Finally,this paper presents a method for the correction of uneven illumination image.Experiments show that this algorithm can effectively eliminate the effect of illumination uniformity,relative to the common local binarization and enhancement method.The recognition rate of OCR has increased significantly,and the combination of global and local methods can improve the scope of application in all kinds of samples and different illumination conditions.