In order to rapidly construct a visual model of rice leaf cells, the authors employed modeling tools of computer aided geometric design (CAGD): curve and surface. Rice mesophyll cells are an object of the study. In virtue of the Bernstein basis function properties and convex combination optimization, this work reconstructs the computation graph of the de Casteljau algorithm to model rice leaf cells, and C++ programming language and OpenGL rendering library are used to establish the visual model of cells. The theoretical results indicate that the control variables of curve and surface shape can be increased by revising affine combination coefficient through the polynomial space of the Bernstein basis function expansion. In addition, the results suggest that the convex combination optimization approach reduces the calculated time complexity of cytoskeleton interpolation points from O(n2) to O(n). The experimental results show that the convex combination algorithm has a significant advantage in terms of computation speed over the de Casteljau algorithm for each 200 interpolation points of the cytoskeleton. In view of the characteristics of close arrangement, regular shape, and large quantity of plant cells, the method proposed in this study has the ability to provide a feasible technical route and a rapid expression for visual modeling of plant cells.