抄録
In this paper, we present an improved semi-automatic model based algorithm for texture replacement of architecture in single image. Compared with previous proposed texture replacement algorithm, the improved one not only enhances replacement accuracy but also reduces amount of user interaction by exploiting relevance between segmented regions. The proposed texture replacement algorithm could help users to picture of architectures with different textures before you paint it as imagine architecture with different textures that you like is a really difficult task. The algorithm proposed is key part of an architecture image processing system developed by us.