Single-valued neutrosophic set (SVNS) is an extension of fuzzy set, which combines the truth, indeterminacy, and falsity information. The measurement of distance of single-valued neutrosophic set will bring new ideas to pattern recognition. This paper introduces the distance definitions and properties of SVNS, and proposes the improved distance definition of single-valued neutrosophic set based on decision maker attitude towards indeterminacy information and applied them to pattern recognition. By assigning a value to decision-maker’s attitude towards indeterminacy information, this attitude comes from the historical experience and risk preference etc. The distance calculated in this way can often get a matching result closer to the reality. The results show that the improved distance formula of SVNSs is more effective and more practical for pattern recognition.