Research Of Ancient Books Electronic Based On Image Processing Technology

In the modern society, ancient books have a great value of economic and culture. Due to various reasons many ancient books suffered from different degrees of damage. We can use digital image processing methods to re-print the books for publishing them. As we all know, old pages are thin and tick. We can choose photoelectric scanner to scan them. To obtain electronic version,we can also use digital camera. Different methods bring different problems, before processing they need classification. Through market research, it was find that there is much demand of image enhancement and feature extraction. There are some ancient books processing software, but they can’t meet the needs of ancient books publishers. A large number of pages are fuzzy, with background, deformation. This paper explored on the existing research results and used different methods to solve the different problems.What the paper did is as follows:1. With the application of different image enhancement methods, we draw a conclusion that:different types of books use different methods, stretching method is good for illustrated pages, the pages with other page background need binary method. Histogram equalization may blur out some of the original image information.2. The paper introduced the algorithm of multidimensional feature vector scale invariant feature transform, looked for the extreme value points to extract the image features in the invariant scale spatial. The experiment results show that this method can only extract the local characteristics of ancient book pages, can not contain the complete structure of the whole pages,and may lose a lot of other information. So this method is not applicable to the feature extraction of ancient books.To a certain extent, it can be used to implement matching.3. The paper also analyzed the invariant feature extraction algorithm based on rank of matrix.Ruled texture image has low rank, low-rank texture contains important structure of the image information. Combined with fast convex optimization method to optimize the rank of matrix to get image invariant features.Experimental results show that under the condition of the projection transformation, the transform invariant low-rank texture method can accurately recover low rank pages, it is an optimal choice for extracting the characteristics of ancient books and correcting deformation ancient books.