In the application of machine vision technology for measurement, the factors of pixel equivalent, system errors and light source intensity in measurement system affect measuring accuracy. In order to improve the measuring accuracy, the pixel equivalent and system errors must be calibrated, and we need to analyze light source intensity which influences edge location of work piece image. An integrated calibration method based on lattice calibration plate in vision measurement system is put forward. On the basis of extracting center coordinates of calibration circle, pixel equivalent is got by calculating the center distance of physical value and pixel value. The error model with the actual coordinates and theoretical coordinates of center point is established, the error model coefficients are got by the least squares fitting. By determining the compensation amount of image edge position error caused by light source intensity, composite calibration of measurement system is achieved. The experimental results show that this method can effectively improve the measurement precision.