A disparity remapping method that enlarges disparities inside objects is described. The method also reduces disparity gaps between objects and background. A disparity map is decomposed into a structure component and a high-frequency component, and two components are manipulated independently. Psychophysical experiments are conducted to evaluate the method. The task of participants in the experiments is to judge whether the object is convex or concave. The result of an experiment using random dot stereogram shows that if disparity gap between object and background exists, the detection threshold of depth structure increased. The result of another experiment using stereo pictures whose disparities are inverted inside objects shows that depth structure inside objects becomes more easily detected by the proposed method.