An Algorithm to Detect Illegal Buildings Using Color Transferring and Texture Difference

Nowadays, the manual inspection of urban buildings still plays a major role in city management. Due to the low efficiency of manual inspection and few researches on illegal building detection based on fixed-point surveillance, it is significant to develop an automatic and high accurate illegal building detection algorithm. In this paper, we propose a novel illegal building detection algorithm which can be applied to a complexity environment. Firstly, due to the local light difference between old and new phase images, we put forward an improved adaptive color transferring algorithm, which is used to obtain the intensity difference image. Secondly, a texture difference image is obtained using the local texture difference algorithm, and noises in the texture difference image are filtered in the HSV color space. Thirdly, we take advantages of architecture characteristics to put forward an improved weight distribution equation for integrating intensity and texture difference images. To verify the effectiveness of our proposed algorithm, we test the accuracy of our algorithm against other three algorithms with 110 pair of multi-temporal images. Visual and quantitative experiments show that the proposed algorithm outperformed other existing algorithms.