Accurate measurement of object perimeters from digital images is one of the important and fundamental tasks of image analysis and computer vision.In this paper

a new accurate and fast perimeter estimator based on eight-connection MSC (m-Sampling and Classification)was presented. Estimation errors were analyzed and predicated .To prove the performance .a comparative experimental study about the exsiting methods and our new method was carried on a digital object datasets including circles with different diameters and rectangles with different edge lengths. Satisfactory results were obtained.