A Novel Approach for Noise Removal from Hand Written Manuscript using Enhanced Gibbs Sampling Algorithm

Hand written documents are one of the most important sources of writing media in ancient Kerala. Due to improper storage of manuscripts, about 80% of them got degraded. Noises are one of the most common degrading factor which affects the visibility of the images and makes them unclear. The main noises found in manuscripts are pepper salt noise, speckle noise and holes on billings etc. There are many denoising methods available. In this paper we have made a literature analysis, both basic and advanced algorithms. Finally compared the output of each methods, Gibbs Sampling technique gives more accurate output in handwritten manuscripts. But holes in documents were not removed. For this we have given an enhancement using Circle Hough Transform(CHT) Which detects the circles and can identify exact hole by comparing the boundary pixels of circle and background of the script. In future we can recognize the characters using Support Vector Machine(SVM) and H2istogram of orientation Gradients(HOG).