In this paper, we propose a new embossing algorithm for gray images using Kalman filter. First, a 2D gray image is first converted to a one dimension vector; those vectors could be considered as a one-dimension discrete-time signal. Then, the performance of image filtering using Kalman filter for image is studied and according to its results, Canny edge detection operators are investigated to find edge map in a gray scale image. Finally, enhance contrast using histogram equalization has been applied. Compared with other conventional embossing method for images, it is an impressive experimental result using our proposed algorithm for gray image embossing. Practical results show that this algorithm can be exploited in different fields such as image pattern recognition.