Analysis of Document Images through Local Binary Patterns

Abstract

Objective: To develop a robust Local Binary Pattern (LBP) method with ENI (Edge pixels, Noisy pixels and Interior pixels) features for enhancement and binarization of ancient document images. Methods: The Local Binary Pattern method is used for improving the image quality by the extraction optimal image features. This method works efficiently for noisy and low contrasted ancient document image. The foreground and background of the low contrasted ancient document images are overlapped and in some documents back ground may be very complex. It is very difficult to separate background from foreground of the document Images. These flaws can be removed effectively by robust Local Binary Pattern method. This method can separate the foreground and background of the ancient document images by enhancing the quality. Findings: Thus the proposed LBP method achieves better results even when more level of noise is added to the image. It enhances and extracts the text in the document image by giving the robust results. The experimental results show that the proposed method produces better results than traditional methods. Applications: The proposed Local Binary Pattern method can be used to improve and extract the optimal image features from the ancient or low contrasted and noisy document images.