The work aims to study the characterization method of scanner image input device based on the least squares support vector machine regression (LSSVR). With the ColorChecker SG standard color card as the target, a nonlinear mapping model of RGB three-channel value to CIE Lab value was established with the least squares support vector machine. The cross-validation grid search was used to determine the optimal parameters of the model and the LSSVR model was optimized to achieve the chromaticity characterization of color scanner. The R-squared of the model's training set was 0.996, the R-squared of the validation set was 0.998, and the average color differences of CIEDE2000 of the training set and the validation set were 1.1463 and 1.2754, respectively. LSSVR model can better realize the chromaticity characterization of color scanners, and has strong generalization ability. It also can effectively improve the characterization accuracy and has a faster calculation processing speed.