Image Mask Fusion for Precise Object Shape Replacement

The patent segments the image processing into distinct components: obtaining a first mask representing the first object's occupied area, obtaining a second mask representing the second object's occupied area, generating a third mask by fusing the first and second masks, and using the third mask to control the shape of the second object in the converted image. This segmentation allows precise control over the second object's shape while maintaining manageable processing complexity.

The patent performs preliminary actions by obtaining and processing masks before the actual image conversion. The first mask and second mask are obtained in advance, and the third mask is generated by fusing these masks before being used to guide the image conversion process. This preliminary preparation of masks enables precise shape control without increasing the complexity of the conversion process itself.

The patent merges the first mask and second mask to generate the third mask, which represents the union of both objects' occupied areas. This merging operation consolidates multiple mask operations into a single fused mask that can be directly used to control the second object's shape in the converted image, improving processing efficiency while maintaining shape precision.

The third mask serves as an intermediary that bridges the first mask and second mask to the final image conversion process. By fusing the first and second masks into the third mask, the patent creates a single control mechanism that enables precise shape control without requiring separate processing steps for each mask, thus maintaining processing efficiency.