Abstarct: Organic petrography is one of the cheapest ways in petroleum industry for primary exploration, which results in reduction of some factors, such as: risk and heavy exploration expenses. These methods are applied by human with the use of microscope, hence there might be a high value of error, despite, by using image processing technique, which is developing so fast in petroleum industry, and the error can be reduced. In this work, by introducing a new algorithm to classify different macerals containing, vitrinite, liptinite and inertinite, we aim to determine each of them percentage, where the kerogen type can be detected. The algorithm includes three techniques containing segmentation technique, which is introduced here for the first time, neural network and discriminant classifier. In addition, to determine the maturity of source rocks by determining the Spore Color Index (SCI), color spectrum simulation in MATLAB network is used. After implementing the code in MATLAB, it is understood that the current algorithm can classify the macerals and also detect the kerogen types with the accuracy of about 62.5% and 70%, respectively. Further, the maturity of source rocks is analyzed qualitatively, where it shows a good accuracy. Meanwhile, the algorithm is validated by experimental data and an expert's observation