| An algorithm for text string separation from mixed text/graphics images is presented. In comparison with those proposed works dealing with extracting text string issues, the algorithm has different features. The union-find operation is employed to perform detection of text regions in digital images efficiently. This scheme is also called region growing.; The principle of this method is to start with some seed objects and then jump from one object to the next. Based on this technique the Euclidean distance which is considered as a distance measure, is needed plus some threshold values in order to define in geometric terms what the next right neighbor should be. Then, we merge these patterns block by block until they are considered to be optimal with maximum neighborhood connectivity. For such a method, the maximal blocks, i.e., potential text regions, are obtained after several union and find operations while the union-find process is considered to be terminated. Accordingly, in terms of image processing or computer vision, such a block labeling has been realized in locating a segment at the same time as the procedure in detecting a text-string region by the union-find operation.; The algorithm is thus able to classify the text from the graphics and adapts to changes in document type, language category (e.g., English, Chinese and Japanese), text font style and size, and text string orientation within images. In addition, it allows a document skew that often occurs in the initial image or the scanned image, without skew correction prior to discrimination while the proposed methods such as projection profile or run length coding are not always suitable for the condition. The method has been tested with a variety of printed documents from different origins with one common set of parameters in this work. The experimental results of the performance of the algorithm in terms of computational efficiency are demonstrated by using several test images from the evaluation.; Consequently, the contribution is primarily to facilitate the achievements of software after extracting text information and graphics interpretation system after separation in terms of fast and effective analysis. |
