To address the issue that the feasible-value constrained scalarization method cannot obtain properly efficient solutions for nonconvex multiobjective optimization problems, this paper uses the idea of model reconstruction and proposes an improved feasible-value-constraint scalarization method, investigating its application in nonconvex multiobjective optimization problems. First, by reconstructing the original multiobjective optimization problem and combining it with the feasible-value constrained scalarization method, an improved feasible-value-constraint scalarization method based on the reconstructed model is constructed. Second, using this method, scalarization characterizations of (weakly) efficient solutions are established for both the original problem and the reconstructed multiobjective problem. Third, based on the relationship between properly efficient solutions for the original problem and (weakly) efficient solutions of the reconstructed problem, a scalarization characterization of properly efficient solutions for the original problem is derived using the reconstruction based feasible-value-constraint method. Finally, numerical experiments demonstrate that, compared with the feasible-value-constraint method, the proposed method can approximate the Pareto front more effectively.