The Application Of Blind Source Separation Algorithm In Scanning Image Processing

With the advancement of our society and the development of information technology, we often need to scan paper documents into electronic documents, in order to facilitate the transmission and storage. But normally, because of semi-translucent of the paper, a scanned image is often the superposition of front and back picture of the paper, that is, the image is the mixed image of two pictures, which reduces the readability of documents. In particular, the machine read or identification, such as optical character recognition, the recognition accuracy will be greatly reduced, thus enhancing scanned image processing is necessary. Obtained by scanning image can be similar to the mixed image which is the original signal as both front and back pictures of the paper linear combination. Using of conventional digital image processing methods is difficult to resolve. To solve this problem, taking into account the basic conditions to correspond the BSS, we can use BSS technology to resolve. The source signals are the two different image signals, taking into account this prior knowledge, to correspond the relevant criterion for BSS, this paper mainly related to select the BSS algorithm which based on decorrelation. Present using the BSS algorithm which based on decorrelation to solve mixed Image separation, so as to achieve the purpose of scanning image enhancement.In this paper, using of MATLAB software, based on the theory of BSS algorithm compile to program which eliminate interference on scanned picture. the program which eliminate interference with the background scanned image principally involves reading the scanned image, image pre-processing, BSS algorithm separation, showing the results images of separation and performance evaluation of several parts. Using the system which elimination of the background scanned picture to achieve separation of the actual scanned image, and evaluate and analyze the results of experiment. The separation results proved that using the system which elimination of the background scanned image to separate the actual scanned image is feasible. Further analysis the results of separation, educe the reasons of scanned images are not completely separate. Experiments prove that the BSS algorithm is applied on deal with the scanned image is feasible and effective. This method can effectively remove the scanned image background interference, improve the identified ratio of the image and text, and achieve the aim of improve the effect of scanned image.