An efficient artistic image-video transfer approach

Abstract

The paper presents a very simple and efficient way of color style transfer of a video using a reference image. Style transfer changes the mood of look and feel of video. Here, in this paper instead of RGB color space, for style transfer oRGB is used which is much more robust compared to other color spaces. Then the color space is segmented using an unsupervised k-means algorithm. Later, each cluster is passed through a histogram matching filter to generate artistic synthetic image. Cumulatively this process generates an artistic video entirely different from the original ones. Several experimental results are shown to prove the efficient of the proposed approach compared to the other existing state-of-the-arts.