In this paper, we propose dead pixel detection and compensation method using nonlinear estimator for infrared camera. Infrared camera has dead pixel that is abnormal output values due to complex factors such as manufacturing process, electronic parts and so on. Dead pixels are able to affect detecting a small target. So, It needs detection and Compensation process. However, after Compensation, some dead pixels are remained and detected by the human. They are soft dead pixel. The key idea of this proposed method, detecting soft dead pixels, is that design a nonlinear estimator using image data characteristics. This propose is able to not only detect soft dead pixels but also pixel Compensation that reflects infrared camera output characteristics well.