Character and Line Figure Recognition in an Image

点击翻译按钮获取中文摘要

The main purpose of this research is to develop a character and a line figure extraction methods applicable to various category of images. As character extraction methods, we have developed a character strings extraction method from magazine cover images and a telop extraction method from news video. The former is based on segmentation by the differential Top-hat operation, and we made effort to establishing the threshold by local histogram method and resolution of the competition between extracted character regions. The latter that is based on the morphological operation realized the detection of the telop frames and extraction the telop regions. For the line figure extraction, we have proposed a detection method of cracks and structural objects of the road surface image. An advantage of the method is its robustness against noise and vagueness of the objects. We have also made study on the parallel computing of the morphology operations and on the distributed constraint satisfaction.