Binarization And Its Application In Preprocessing Of Cheque Recognition System

Binarization is a common technique used for segmenting image. The essence of binarization is using a threshold of gray level to classify the pixels in an image. As an important image segmenting technique, binarization has great impact on the final recognition result of image recognition system. A image recognition system usually has two subsystems: preprocessing subsystem and pattern recognition subsystem. The preprocessing subsystem has to segment the target and extract characteristics from the target.Binarization is the main method to segment target. No matter how well the recognition algorithm is, image recognition system can not finish its recognition task correctly, if the target is poorly segmented. Check recognition system uses DIA (Document Image Analysis) and OCR (Optical Character Recognition) techniques to process the check images. Check recognition system has great academical values and commercial values and has being studied by many scholars recently. In the check recognition system, a check firstly scanned by specially designed scanner in order to getting a digitalized image. Then, the critical step of the check image preprocessing is binarization. As a foundation for the further process, binarization extracting the interesting objects from the whole image has huge impact on the check recognition system.This paper studies some of the binarization algorithms based on the references [9][10][11][12][13][14]. After analyzing the histograms of gray images of 2000 cheques, we found a clue, which can be used for segmenting image. With the clue, we developed a new binarization algorithm based on the maximum gradient of the histogram, making the target object stand out in the binary image. Comparing several other commonly used binarization algorithms, the algorithm proposed by the paper has been proven in the simulating tests to be more feasible and advanced. The contributions of this paper are: Concluded three different kinds of check image processing patterns. Introduced and studied several binarization algorithms. Introduced the results of line-detecting and some other measurements to evaluate the quality of binarization. Studied the relationship between the scanning resolution and image. The innovative works of this paper are: (1) Proposed a new and effective and reliable algorithm based on the maximum gradient of the histogram of the image for check image binarization. (2) Introduced the results of line-detecting to evaluate the quality of binarization.