Uyghur Text Detection in Natural Scene Images
Xinming Li, Junfang Li, Qiang Gao, Xiao Yi Yu · 2019
Text detection in natural scene images is a challenging task for many content-based image analysis missions. Actually, widely-used methods mainly focus on commonly used languages, while for the Uyghur language of minority, text detection is paid less attention. In this paper, we propose an effective and accurate Uyghur text detection method based on Sobel edge detection algorithm. Binarization, morphology operation and connected components analysis are used to detect Uyghur text regions successively. In addition, location of Uyghur text was realized by text regions merging. The proposed algorithm is evaluated on 4 representative groups of natural scene images. The True Positive Rate (TPR) of each picture is over 0.78 by our method which is better than the Maximally Stable Extremal Regions (MSER) method and the texture based method. Experimental results show that for natural scene images, Uyghur text detection and location can be obtained by proposed method.