根据映射方向的不同，传统Image Warping算法可分为正向和逆向方法。针对正向方法存在的映射出界和扩展空洞问题，以及逆向方法存在的效率瓶颈，设计了一种双向缩放Image Warping算法。新算法将传统的单向映射分解为一个逆向的视景收缩和正向的模板映射过程，仅通过调整目的图像绘制顺序和映射方向来进行改进，而避免了其他优化算法中相应的预处理步骤以及对场景本身的限制。理论和相关实验证实了双向缩放Image Warping算法的合理性和有效性。

Image Warping includes the forward and backward methods respectively according to the mapping directions. However, out-mapping and holes often exist in the forward approaches, while backward ones cost much computing resources. A novel bidirectional zooming image warping method is proposed in this paper. Instead of the traditional unidirectional mapping, a backward mapping and subsequently a forward mapping are performed to achieve the warping. The new method gives a good solution to current problems by only adjusting the direction of mapping and the order of drawing, without any extra preprocessing and limitation of scene. The promising results can be obtained from both theoretical and experimental studies.