Abstarct: Properties of reservoir play a significant role in management and development of hydrocarbon reservoirs. Permeability is the ability of porous rock to transmit fluids and one of the most important properties of reservoir rock, because hydrocarbon production depends on reservoir permeability. This parameter can be measured at the laboratory utilizing air injection analysis, or by using well test. These methods are expensive and time consuming. In recent years, the capabilities of intelligent systems and image analysis has been proven by development of hardware and software of computer. Therefore, image analysis and intelligent systems has been used for permeability prediction to reduce time and money. The current study includes two main parts. In the first part, the permeability was predicted from petrographic image analysis and intelligent systems. The thin sections of well A in south pars gas field were used. Petrographic image analysis was employed to measure the types of porosity including intergranular, intragranular, moldic, micro and optical, amount of cement, calcite, dolomite and anhydrite, type of texture and mean geometrical shape coefficient of pores. Three intelligent systems including shallow neural network (NN), fuzzy logic (FL) and neuro fuzzy (NF) were used to predict permeability. The permeability was predicted using the three individual intelligent systems and petrographic features. The mean squared error (MSE) of the NN, FL and NF methods in the normalized test data are 0.0107, 0.0081 and 0.0080, respectively. The mean absolute error (MAE) of the NN, FL and NF methods in the real test data are 11.54, 11.17 and 10.85 mD, which correspond to the R values of 0.9397, 0.9588 and 0.9558, respectively. The concept of committee machine was used to improve the accuracy of prediction. Hence, two types of committee machine with intelligent systems (CMIS) were used to combine the predicted values of permeability from the individual intelligent systems: simple averaging (SA) and weighted averaging (WA). In the weighted averaging, a particle swarm optimization (PSO) was employed to obtain the optimal contribution of each intelligent system. The MSE of CMIS-SA and CMIS-WA in the normalized test data are 0.0072 and 0.0066, respectively. The MAE of CMIS-SA and CMIS-WA in the real test data are 9.03 and 8.94 mD, which correspond to the R values of 0.9623 and 0.9622, respectively. Comparing the results of the first step with core data, shows that petrographic image analysis and intelligent systems were successfully applied in permeability prediction. Among the intelligent systems, the committee machine with intelligent systems obtained the most accurate results. In the second part, the NMR permeability was compared with the results of the first step. The NMR permeability of well A was unavailable. Therefore, a deep neural network was constructed with the information of well B including GR, NPHI, PEF and RHOB logs as inputs and NMR permeability as output. The MSE of the deep neural network in the normalized training data are 0.0020. This model was used for prediction of NMR permeability in well A. So, the information of well A including GR, NPHI, PEF and RHOB logs were input to the model and NMR permeability was predicted. The core permeability and predicted NMR permeability versus depth were plotted and the correlation coefficient was calculated. The R value of 0.9767 shows that the predicted NMR permeability have a good correlation with core permeability. So, the predicted permeability from image analysis and intelligent systems were compared with predicted NMR permeability. Comparing the results of the first step with predicted NMR permeability, shows that the integration of petrographic data and intelligent systems performed well.