Abstract: In this paper a method for measuring crack dimensions was introduced based on machine vision. this method was used because of the irregularity of the crack shape and difficulty of measuring crack dimensions by the traditional ways, A combination of operations were used to calculate surface crack length after filtering the binary image obtained from captured one, the key factor for best results is using the correct order of the operations on the image. This done by MATLAB image processing tool box. A proposed formula was introduced to improve the accuracy of the crack measurement. The accuracy of the measurement greatly depends on the resolution of the captured image and camera type. Many samples were measured and the results showed the error percentage in this method fluctuates between 0.3%-1.9%