The calculation method of color difference in images is still an unsolved issue. A series of experiments and calculations have been done to test the consistency between computing color difference and perceptual one in images. Five ISO SCID images N2, N3, N4, N5 and N7 were used as the test image in the experiments, whose color were altered in lightness, chroma and hue independently or simultaneously to form the test image pairs. Each image pair was of different color difference grade. CIELAB, CIE94, CIEDE2000 and CMC color difference were computed by averaging color difference pixel by pixel for digital images and by averaging color difference of 256 typical color patches extracted from each image for printed images. The digital test images were displayed on an EIZO CG19 LCD and the printed test image pairs were viewed in a D50 light booth. The experimental results showed that the lightness, chroma and hue difference behaved differently when plotted the perceptual color difference against computed ones. This implied that the color difference formulae should be optimized and different weighting factors should be added to different visual attributes. The color difference formulae can be optimized by the slope of fitting lines of C*/L*. The optimized CIELAB(1.5:1), CIE94(3:1), CIEDE2000(2.3:1) and CMC(3.4:1) for digital images and CIELAB(1.58:1), CIEDE2000(1.75:1) and CMC(1.84:1) or CIELAB(2.44:1.54:1), CIEDE2000(2.76:1.58:1) and CMC(2.74:1.49:1) when hue is considered for printed images by experiment data performed much better than the original formulae.