The division of the content from the recorded debased picture reports is an extremely difficult undertaking on the grounds that the varieties between the closer view content and foundation content are difficult to decide. In this paper we have actualized canny edge detection algorithm to examine and remove the words from a corrupted picture, the procedure conveyed here is the picture contrast which is adaptively found to settle the issue. At first the differentiation guide is taken from the debased record pictures. The mix of nearby picture angle and the neighborhood picture difference is the versatile picture differentiation, and afterward it is changed over to double level and joined with canny edge detection algorithm and artificial neural network to concentrate content edge pixels. The record content is further portioned by a nearby limit that is evaluated in view of the powers of distinguished content stroke edge pixels inside a neighborhood window.