Abstract：Binarization for Color Caption is one of major tasks in Caption Recognition. Because of the limitation of the widely adopted threshold method for caption binarization, especially for color caption, we propose a binarization method for color caption based on Fuzzy Histogram. As there is great difference between the foreground pixel and the background pixel in color value, this method classifies all the pixels according to the membership degree of the Fuzzy Histogram by color value, then it sorts the classes by number of pixels in each class and picks the top 20 classes, after that, it selects the class in which the pixels are all belong to the foreground using the inherent characteristics of caption, such as the width of stroke, the ratio of foreground and background and so on, thus completes the binarization of caption image. The experiment results show that this method not only remains fast speed but also has a good effect on the binarization of caption image with complex background and is much better than threshold method.
Kun Liu;Xueqiang Lv;Tao Cheng;Shuicai Shi. Binarization for Color Caption Based on Fuzzy Histogram[J]. , 2009, 6(3): 1217-1222.
Kun Liu;Xueqiang Lv;Tao Cheng;Shuicai Shi. Binarization for Color Caption Based on Fuzzy Histogram. , 2009, 6(3): 1217-1222.
2009, Vol. 6 
