Abstract:For the Chinese text location under complex background, this paper presents a novel method by combining Gabor filter and support vector machine (SVM). It bases on such a fact that Chinese characters are composed of four kinds of strokes. By extracting four kinds of stroke features with Gabor filters, Chinese text location problem can be transformed into a texture classification one, which can use SVM classifier for the purpose. So, the proposed method is composed of two phases. First, Gabor filters with different scales and orientations are employed to obtain four texture images representing the stokes of Chinese text in horizontal line, top-down vertical line, left-downward slope line and short pausing stroke directions. Then, the text regions and background regions in four texture images are used to train four SVM classifiers to distinguish the texture in four directions, by integrating an SVM classification network to obtain the final classification results, according to the sum of the weights to determine whether the block is the text region. Some experiments are conducted on a large amount of typical images with different texts and different fonts. Compared with some existing methods, the proposed approach achieves better results for Chinese text location.

Extraction of text under complex background using wavelet transform and support vector machine

A Novel Approach to Text Detection and Extraction from Videos by Discriminative Features and Density

Method for text region localization in complex background images

Text Extraction from Mail Images with Complex Background.

Chinese text location under complex background using Gabor filter and SVM

An Automatic Video Text Detection, Localization and Extraction Approach.

Texture recognition with support vector machines

Text Detection Based On Wavelet Transform

An Automatic System for Text Location and Extraction in Digital Video Based Using SVM

Fast and robust text detection in images and video frames

Texture Classification Based on Support Vector Machine and Wavelet Transform

Textural image segmentation with multi-scale wavelet analysis based on feature learning

Text Extraction from Complex Background Images

Caption Location and Extraction in Digital Video Based on SVM

Text Detection Based on Stroke Features

Texture Segmentation Using Kernel-based Techniques

Character Segmentation Method of Check under Complex Background

Image steganalysis based on multi-feature extraction using support vector machines

New method for text detection and segmentation from complex images

Research of Text Area Extraction from Color Images with Complex Background

Face Recognition Based on Wavelet Decomposition and MultiSVM