An experiment was conducted to study the effects of working parameters on performance indexes of a new-type vertical plate seed-metering device. The regression model of qualification indexes, missing seeding indexes, variation rates of seedling distances and experimental factors was obtained with multiple-factor and multiple-level experimental design by using computer vision technology and professional data processing software. Results showed that plant-space is most important on influencing performance indexes among the experimental factors, and the good level of arbitrary two factors combination could remedy the bad level of the third factor, the level in pairs or groups of the factors could be arranged to suit different working and corresponding indexes demand. 