Due to the history of culture and technology,a large number of Mongolian documents are in the form of images stored in the library. Documents in image format take time and effort in search and retrieval,extremely inconvenient. In order to solve this problem,this paper proposes a method of locating Mongolian words based on convolution. The two- dimensional discrete fourier transformation model,the two- dimensional discrete convolution model and the transformation of the two are given. Applying the above mode...