Skew detection via principal components analysis

Abstract

Skew detection via principal components is proposed as an e ective method for images which contain other parts than text. It is shown that the negative of the image leads to much more robust results, and that the computation time involved is still practical.

Hough transform based fast skew detection and accurate skew correction methods

“…We compare the accuracy and speed of skew detection by the proposed method with some other methods given in the literature [16][17][18][19][20]. The method of Steinherz et al [16] is based on the projection profile and principal component analysis (PCA).…”

“…The method of Steinherz et al [16] is based on the projection profile and principal component analysis (PCA). The method is very fast as its complexity is O (N) where N is the number of image pixels.…”

Time and Accuracy Analysis of Skew Detection Methods for Document Images

“…As mentioned in "Ref. [12,13]" for a given binary image, firstly black pixels of the image are mapped in a two dimensional vector. Then, the unit vector for the maximized projection profile deviation is obtained.…”

“…4. Using Radon function, find the Radon transform for theta ranging from 0 to 179° using " (12),". 5.…”

Skew Detection in Handwritten Documents

“…Hinds et al [1] uses run-length coding and Hough transformation while Cao [2] uses line-fitting followed by Hough transformation for detecting skew.Other methods for skew detection include cross correlation used by Chen et al [3], projection profile by Steinherz [4], Fourier transformation by W. Postl [5]. All of these methods give high accuracy but none of them focus on the memory constraints involved while processing the documents.…”