Detecting corners of text in spam images
Wan Ming-cheng, Geng Ji, Hongrong Cheng, Zeng Zhihua · Computer Engineering and Applications Journal · 2009
Combining edge extraction methods and a circular template,this paper proposes a new method for detecting corners of text in spam images.Firstly,the edges of text in spam images are extracted by color-roberts algorithm and threshold segmentation. Then,the corners is detected by using a circular template.Most noise in spam image is eliminated by edge extraction and threshold segmentation.Besides,the novel circular template allows the corner detection is insensitive to the orientation of text.A comparative experiment engages SUSAN method and the proposed method for the corners detection of text in images from a real spam image archive.The results indicate that the method has a better performance and can obtain the angle magnitude of corners contemporarily.