LOCATING TEXT IN IMAGES BASED ON THE SMOOTH GRAY-LEVEL DETECTION
Shiueng-Bien Yang, Chi-Feng Wu · International Journal of Image and Graphics · 2012
The morphological technique is widely used for text extraction from images. It is highly effective in extracting geometrical structures and representing shapes in images. However, the morphological technique does not consider the variance of gray levels between neighboring pixels in images. Generally in images, changes in gray levels between neighboring nontext pixels are smoother than those between neighboring text pixels. This paper, therefore, proposes a new method for locating text based on the smooth gray-level detection. Experimental results reveal that the proposed method locates text in images better than other methods.