Image stacks acquired by TIRF microscopy system suffer intensity loss due to photobleaching with time. In this paper, a new attenuation compensation method for TIRF microscopy images is proposed. Loss of intensity is assumed to vary only along time and governed by an exponential decay law and the decay process is then modeled as a decay function. The parameters of the decay model are estimated by iterative reweighed least-squares method. In order to excluding the contribution of the outliers, only foreground pixels are participating in parameters estimation. Finally, the restored image is wrapped backward to reference frame. The algorithm is applied in TIRFM stacks containing GFP labeled GLUT4 vesicles and the experimental results indicate that the corrected images stack are in more stable brightness along time.