A New Method for Image Cartoonization Based on Mean-shift and DoG
Huang Sheng-xue · Coal Technology · 2009
It is very useful to convert digital images into cartoonized style in order to conveniently produce cartoon films.In this paper,we propose a new method to generate cartoonized image(also called image stylization) from digital images.Firstly,a mean-shift filter is used to get rid of the color and texture details such that the regions with same attributes are replaced by larger color blocks.Secondly,the filtered image is segmented by Difference of Gaussian(DoG) filter to extract the edge information.Finally,the filtered image and the edge map are fused together to get the cartoonized image.Experimental results validate the effectiveness of new method.