A novel method for slant correction of Persian handwritten digits and words

In this paper, a new slant estimation/correction method for Persian handwritten digits and words is presented. Based on the presented method, first using original image and its inversion, a new combined image is constructed. This new image has the same slant as the original image. So its slant is estimated and considered as the slant of original image. Finally, using a shear transformation, the slant of the original image is corrected. The performance of the presented method has been tested using several popular Persian/Arabic databases. Obtained experimental results and comparison with the other methods in the literature confirm that the presented method is accurate, and promising.