Extraction Method Of The Seal's Image Archive Chinese

There are a wide variety of information in image. It is the focus of attention of domestic and foreign scholars how to extracted interesting information from complex background images in quickly and accurately. Extraction of text positioning is a key step in image text recognition. It has a direct impact on the results of the text distinguishably system.On the basis of the key different point of text extracting,there are some different problem what is resolved. there is no a common standard methods and evaluation criterion on the algorithm of text location.Utilizing image text location technology in present-day,aim at the feature of bill text,Researching and improving some core technology of the notes text preprocessing.In the normal circumstances,there are many form lines overlapping with text in the color bill.we can locating two lins edge using Hough in this paper accurately. Judging eight-neighborhood of one pixel to Remove obstruction of form lines.Extraction and location algorithm of the printed text are key research in bill image.Combine with different edge detection operator,we can find the bill text edge.After morphological operations,there are many simply connected region of similar text feature.And can remove small area noise.Improve the traditional bilinear interpolation method,the single character can remain the original strokes. Make Use of projection method to split a single binary character, the enlarged character can be recognized from OCR after madial axis transform algorithm.In order to remove dark spots and printing text of the bills in the background,In this paper,the RGB color components transforming is Utilized to separate red seal from background.Then we preprocessing the seal by means of linearity transform to strengthen gray image,Otus, mathematical morphological operation. Finally, the good binary seal image will be obtained.