Contrast enhancement for images of raised characters on region of interest

Raised or indented characters are widely used for recording the product information in the industrial field. However, images of such characters are of poor contrast because the characters have the same color and material as the background. It is this problem that makes recognition of such characters is difficult than that of conventional optical characters. To address this issue, the aim of this work is to investigate a contrast enhancement method applicable for raised or indented characters by carrying out an independent study on the illumination factor. Firstly, the illumination model of raised character is analyzed. Then, four different algorithms are introduced and the experiment was carried out with images of the raised characters pressed on metal label. Experiment results prove that homomorphic filtering is the most suitable contrast enhancement algorithm for raised or indented characters.