Retinex-Guided Channel Grouping-Based Patch Swap for Arbitrary Style Transfer

Since the finite features harvested from one single esthetic style image are inadequate to represent the rich textures of the content natural image, existing techniques treat the full channel style feature patches as simple signal tensors and create new style feature patches via signal-level fusion, which ignore the implicit diversities existing in style features and thus fail in generating better stylized results. In this article, we propose a retinex theory guided, channel grouping-based patch swap technique to solve the aforementioned challenges. A channel grouping strategy groups the style feature maps into surface and texture channels, which prevents the winner-takes-all problem. Retinex theory-based decomposition controls a more stable channel code rate generation. In addition, we provide a complementary fusion and multiscale generation strategy to prevent unexpected black areas and over-stylized results, respectively. Experimental results demonstrate that the proposed method outperforms the existing techniques in providing more style-consistent textures while keeping the content fidelity.