Techniques for reducing interference in images

Techniques for reducing interferent objects in a first image are presented herein. The system may access the mask and the first image. Interferent objects in the first image may be within the region of interest and may have pixels with initial attributes. In addition, the system may process the first image and the mask using the machine-learned repair model to generate a repair image. Pixels of an interferent object in a repair image may have repair attributes in a chroma channel. Further, the system may determine a palette transform based on a comparison of the first image and the restored image. The transformation attribute may be different from the repair attribute. Further, the system may process the first image to generate a recolored image. Pixels in the recolored image may have recolored properties that are based on transformation properties of the palette transformation. 在本文呈现用于减少第一图像中的干扰物对象的技术。系统可以访问掩膜和第一图像。第一图像中的干扰物对象可以在感兴趣区域内，并且可以具有带有初始属性的像素。另外，该系统可以使用经机器学习的修复模型处理第一图像和掩膜，以生成修复图像。修复图像中的干扰物对象的像素可以在色度