This paper attempts to consider the results of a magnetotelluric (MT) geophysical survey and the applicability of image processing method on the TM images of the anticline. The color pixels related to the salt zones have been separated from the other zones from the land surface using image processing method incorporating k-means clustering approach as one of the simplest unsupervised learning algorithms. Field investigations, geological study and the results obtained from the MT survey together with the image processing method identified three salt zones in the study area. These salt zones match well with a decrease in the resistivity of MT data. The application of the proposed method as one of the artificial intelligence technique can be properly used as a cost effective method to identify salt zones from the other areas and preparing a probability map in a regional scale in petroleum exploration.