Image Coding for Binary Document Based on the Regional Features

Lossless coding is commonly found in binary image encoding with lower compression ratio. In this paper, image segmentation is used to classify the document image into line image regions, text image regions and halftone image regions. According to different features of each region, different encoding methods are applied to improve the image compression ratio. Adaptive arithmetic coding is used for line image regions, while symbols dictionary encoding for text image regions and the vector quantization coding for halftone image regions. Experiments show that this method can effectively improve the compression ratio of binary document image.