We study the support vector machine (SVM) classification problem of interval numbers, define the mapping relationship between interval numbers and the hypercube, and build a hypercube representation frame based on interval number samples. We then propose a super planar vertex sampling method based on the complete traversal of the binary tree, which can satisfy sample constraint. We also build up a classification-learning model through transforming the classification objective function. Experimental simulations verify the feasibility and effectiveness of the proposed method.