Stream Flow Painting from Real Images Using Anisotropic Band-Pass Filter

Xiaohua Zhang
2014 Science Journal of Circuits Systems and Signal Processing  
A new non-photorealistic rendering algorithm is proposed for creating artistic painting with soft stream flow from natural color images. The algorithm consists mainly of two stages, that is, a revised bilateral filter called trilateral filter is firstly applied to original color image for creating drawings using gradient information and then a DoG-like band-pass filter is adapted for generating soft stream flow along the eigenvectors and therefore the image is smoothed along curved stream
more » ... The proposed trilateral filter is an extension of bilateral filter by incorporating gradient space. On the other hand, DoG-like band-pass filter is designed by applying eigenvectors and eigenvalues of a structure tensor matrix calculated at each pixel. Our approach effectively preserves image main structures while smoothing image regions in an anisotropic way. Even in regions with lower contrast, stream flow-like potential structures are also well produced due to a gradient relaxation. The experiments demonstrate that the proposed algorithm works well and produces good and pleasant visual results.
doi:10.11648/j.cssp.s.2014030601.15 fatcat:cgt3ku747vdnfimvp3n4mo2q2q