Layout Analysis Based on Multi-feature and SVM
Pla Information · Shuju caiji yu chuli · 2008
The gray feature,the shape feature,and the texture feature are analyzed.The differences among the text,the table,the figure,and the image on document images are studied.Aimed at the Chinese/English layout,the projection profile and the connected component analyses are effectively combined to segment images.Then,the feature vector of seventeen dimensions is abstracted for representing block information,and SVM is introduced to classify document images as four types(text,table,figure,and image).Experimental results show that the method can segment document images for various kinds of the layout segmentation.