Granularity is one of the most important visual artifacts to evaluate the performance of an imaging system, including a printer. At present, granularity-measuring techniques can only be applied on patches with uniform color. However, human can easily perceive graininess on any image; Hence, in this paper, we propose an algorithm to quantify image graininess. First, a flatbed scanner is adopted for its capability to capture large image area. A short-time Fourier transform technique is devised to accurately remove color screen halftone signal without any prior knowledge of the adopted screen signals. At last, users can pinpoint the area of interest to measure color graininess based on CIEDE-2000.