A new fast approach for skew estimation using moments and wavelet transform

In this paper, a new technique based on moments and wavelet transform is proposed to determine skew angle of document image. This technique is divided into two parts. Firstly, the size of image is reduced using two-dimensional Discrete Wavelet Transform (DWT). Then the moment based method is applied on reduced image to determine skew angle. Hence, proposed algorithm has better time complexity and less computational cost than the existing algorithms. This method is tested over large number of documents having skew between -90° and +90° and is observed to be simple, fast, accurate and works with most of the languages. It is also found to be robust against noise Results of proposed algorithm are compared with moment based approach and other commonly used existing methods and found to be to the best method.