Inferring dimensions from an image using automatic calibration

We propose a new technique to infer the dimensions of a real object using an image via the automatic calibration of a camera. This method utilises different geometric shapes to make comparisons for calibration and thus decrease the number of errors. Various square, rectangular and triangular shapes are used as the origin by which to calibrate the camera. Since the position from one of the geometrical shapes is known, thus measurement can be inferred for the objects or plane in with it lies. In this paper we provide several ways to calibrate the camera leading to a higher accuracy of measurement.