Gray-Scale Morphology Based Image Segmentation and Character Extraction Using SVM

Signs and notices are widely used for finding public places and other locations. However, information on signs or notices is inaccessible to many visually impaired people. Therefore, automatically reading text from natural scene images becomes an important application to assist the visually impaired. However, finding text in scene images is a great challenge because it cannot be assumed that the acquired image contains only characters. Natural scene images usually contain diverse text in different size, fonts, orientations and colors, and complex backgrounds such as windows, bricks, and character-like texture. Therefore, this paper proposes a new method to support the scene text reading. This method mainly includes tow parts: (1) image segmentation and, (2) character extraction. The algorithm is implemented and evaluated using a set of natural scene images. Accuracy of the proposed method are calculated and analyzed to determine the success and limitations. Recommendations for improvements are given based on the results.