Novel Approach for Image Text Skew Detection and Correction using Hough Transform
Priyanka Yadav, Ritam Dutta · 2024
As we scan the text images using digital photography devices like cameras, scanners etc. in general, there has always been an angle displacement occurs from the original axis of the text image. This skew of the image(s) often makes us uncomfortable to visualize the text in the image and even when we use such skewed text image with optical character reader, as it degrades the resulting accuracy. Therefore, skew detection and correction is the primary step to design a proper optical character reader, whose primary task is edge-detection and skew in text image using Hough transformation. The proposed algorithm is developed for detection the skew between -1° to +180° angle variation and it has been tested over 25000 distinct text images for testing and the state of the art accuracy is recorded $98.32 \%$ which is considerably better than existing methods.