PixelShuffler: A Simple Image Translation through Pixel Rearrangement
Omar Zamzam · 2025
Image-to-image translation is a topic in computer vision that has a vast range of use cases ranging from medical image translation to image colorization, super-resolution, and generating photorealistic images from sketches. Image style transfer is also a widely researched application of image-to-image translation, where the goal is to synthesize an image that combines the content of one image with the style of another. Existing state-of-the-art methods rely on complex neural networks to achieve high-quality style transfer, but they can be computationally expensive and complicated to implement. In this paper, we propose a novel pixel shuffle method that addresses the image-to-image translation problem generally with a specific demonstrative application in style transfer. The proposed method shuffles the pixels of the style image to maximize the mutual information between the shuffled image and the content image. This approach preserves the colors of the style image while maintaining the structural details of the content image. We demonstrate that this simple method produces comparable results to state-of-the-art techniques, as measured by the LPIPS loss and the FID score, while having significantly reduced complexity, offering a promising alternative for efficient image style transfer, and a promise in usability in general image-to-image translation tasks.1