In order to obtain the optimal threshold for automatic image segmentation efficiently, a novel image thresholding method is proposed based on the conception of Tsallis generalized divergence. First, the original formula of Tsallis generalized divergence was simplified, and then the symmetrical version was constructed. Subsequently, the divergence sum of the foreground and background between original and thresholded image was set up based on the symmetric equation. Finally, the sum formula was minimized to obtain the optimal threshold. The experimental results show that the proposed method is feasible and has better adaptability on complicated images. So, it is an available threshold selection method.