关键词: 红外图像分割, 差分进化算法, Tsallis 熵, q 值, 尺度函数

Abstract: It is easy to lose edge information when using the traditional segmentation algorithm, which leads to wrong judge of the damage area size. Aiming at the infrared image segmentation of aircraft skin damage, a segmentation method combining 2D Tsallis entropy multi-threshold and improved DE algorithm is proposed. Firstly, the initial population number, i.e. the number of thresholds is determined, calculating the q value of each region by using the threshold value. Then the initial threshold mutation and intersection are conducted, selecting the optimal threshold according to the Tsallis entropy value, and the optimal threshold is used as the initial threshold of the
next differential evolution. The optimal threshold is used to calculate the scaling function F of differential evolution algorithm, repeating the differential evolution until the global optimal threshold is obtained.Experiments show that compared with traditional methods, the current algorithm can effectively improve the segmentation efficiency as well as the quality of segmentation results.

Key words: infrared image segmentation, differential evolution algorithm, Tsallis entropy, q value, scaling function