Abstarct: Significant percentage of hydrocarbon reservoirs in Iran, are of carbonate type. One of important features of these carbonate reservoirs are natural fractures which plays a main role in migration and production of hydrocarbons. Consequently, fracture identification in different aspects of reservoir development and management is a vital practice. Borehole image logs are powerful tools for fracture study in drilled wells, which is perfectly competent in presenting high resolution “pseudo picture” of the borehole wall. Image logs with their specific technology, by implementing electrical and sonic characteristics of the formation, are capable of recognizing a wide variety of features in the borehole wall including: fracture, condition of fractures whether mineral-filled or open, faults, bedding, sedimentary structures and etc. In this research, fracture identification using FMI data from two boreholes A and B, in one of hydrocarbon reservoirs located in southwest of Iran, is presented. For this purpose, two methods were utilized to studying fractures. In the first method, close and open fractures were highly and accurately identified and classified using integration within results of Geolog software, manual interpretation and well logging. 262 open and close fractures with northeast – southwest general strike in borehole A, and 62 open and close fractures with east – west general strike in borehole B were determined. Highest density and aperture of fractures zones were determined by density and aperture of fractures in two wells.In the second method, by considering the fractures of FMI images as an edge, the sobel filter, which is one of the powerful filters for detecting the edge in image processing, was employed to study its performance in FMI images and it was indicated that only the results of open fractures detection are acceptable. Generally it can be concluded that in this algorithm, reliable results can be obtained only when the fracture has high physical contrast with its background or in other words, only when the resistivity difference between fracture and background is high.