ICDAR '03 Proceedings of the Seventh International Conference on Document Analysis and Recognition - Volume 2
Texture for Script Identification
IEEE Transactions on Pattern Analysis and Machine Intelligence
Hi-index | 0.00 |
Abstract: In this paper, we propose a novel approach to determine the positions of text areas in images with complex-background using wavelet decomposition and pseudo-motion of images. In our method, a 拢xed image is translated and provides a sequence of moving images-the pseudo-motion sequence of image. In fact, we can consider the translation of an image to be a motion of eyeshot. When an image is translated, its wavelet coef拢cients will oscillate. From this property, we can locate the text areas in complex-background images. Experiments were conducted to demonstrate the performance of the method. In the experiments, we detect the text areas of several different type of characters in images with multi-gray level complex back-ground.