Grain-size distribution or aggregate gradation is one of the key factors affecting the various branches of civil engineering such as geotechnical engineering, river hydraulics and mechanics of sediment transport, material engineering and concrete technology. Sieve analysis (mechanical gradation) of aggregates that is performed using standard sieves needs a lot of cost and time. Therefore, the use of novel methods such as image processing is important and undeniable in the development of industries that deal with aggregate gradation. In this paper, an efficient procedure is suggested to determine the particle size distribution based on a new parameter called self-adaptive size parameter. This parameter is minimum dimension of the equivalent shape of a particle that is calculated by the proposed algorithm. In present research, two aggregate samples with different gradations were employed to demonstrate the efficiency and accuracy of the proposed procedure. For this purpose, the images of the aggregate samples obtained from Aji Chay River, located in the northwest of Iran, were taken using an 8MP camera. The resulting images were then processed using Geological Image Analysis Software and the proposed algorithm as a MATLAB-based program. The study of the output graphs indicates that a good agreement exists between the results from sieve analysis and those from the theoretical gradation based on the proposed procedure.