Digital images are playing a key role but while transmitting the image more disturbances are produced by the noise which corrupts the image. Denoising leads to good quality image and restoration of original information. To achieve denoising, various noise models are referred based on additive and multiplicative type also. Some are Gaussian noise, salt & pepper noise, speckle noise and Quantization noise. The filters are performing well in removing noise that is impulsive in nature. Denoising of images are done using the linear and nonlinear filtering techniques. The linear filtering is achieved using the mean filter and the LMS adaptive filter while the nonlinear filtering is performed using median filter. In this paper, a robust two state filtering technique is presented to remove Salt & Pepper noise in Grayscale image. First step, the algorithm identifies the center pixel and checks whether it is corrupted pixel or not. Second step, it will identify the surrounding pixel and it checks for corrupted pixels in the selected window, the corrupted pixels are replaced by the estimated value using the proposed filtering technique. The proposed filtering technique will produce better results compared with standard median filter.