Automatic text skew estimation in document images

This paper describea an algorilhm to estimate ihe tezt skew angle in a document image. The algo-rithm utilizes the recursive morphological transforms and yield8 accurate estimates of tezl skew angles on a large document image data set. The algorithm computes the optimal pcrrameter set-tings on the fly without any human interaction. In 2hM automatic mode, ezperimental resulti indicate that the algorilhm generates estimated tezt skew angles within t& $ of the true tezt skew anglea with a probabilily of Fo process a So&lpi document image, the algorithm takes 10 seconds on SUN spare 10 machines