Aiming at the problem that one-dimensional generalized fuzzy entropy thresholding failed to segment the noised image, a two-dimensional membership function on the two-dimensional histogram is defined and a two-dimensional generalized fuzzy entropy thresholding method is presented. The proposed method segments an image with the pixel intensity and with the pixel′s local average gray value, which considered more image information. In order to improve the speed and select suitable parameter, a nesting optimal searching process is designed with particle swarm optimization for two-dimensional generalized fuzzy entropy thresholding algorithm. Experiment results show that two-dimensional generalized fuzzy entropy thresholding method has good adaptability with noised images.