Flue-cured tobacco grouping is the basis for automatic grade recognition of tobacco acquisition quality. This paper presents an ant colony optimization that can be used in flue-cured tobacco grouping identification. Compared with other algorithms, it can be used to identify tobacco grouping more quickly. Firstly, collect tobacco from three different locations and of three different colors, and there are 20 samples respectively in each part location and each color. And then extract the image and spectral characteristics of each sample, altogether 16 parameters. Randomly classify the 140 sets of samples. Finally, achieve location and color automatic grouping of the samples by ant colony optimization. The results revealed that, compared with the original grouping, the location and color classification rate based on ant colony optimization is as high as 91.7%. This suggests that the use of ant colony optimization is feasible for classifying unknown tobacco groups