针对现有的多层抗锯齿技术在成像深度方向陡峭的三维(3D)图像时出现深度不连续的问题，提出了一种基于分类映射的真三维深度抗锯齿算法。鉴于固态体积式真三维成像是三维曲面的特点，通过分析YOZ（或XOZ）平面截取三维曲面所生成的二维曲线，提出了符合人类立体视觉特性的、基于相邻多层映射的体素分解方法，以及体素映射类型的判定算法。通过原始体素的自适应分类映射，可以在一定程度上扩大深度连续条件下的偏轴观看角度。在搭建的固态体积式真三维立体显示系统上测试三维成像，在较大的偏轴观看角度（约45°）下多数三维成像在深度方向过渡平滑，可以获得较好的深度连续性。

To solve the problem of depth discontinuity of three-dimensional (3D) images having steep surface on imaging depth direction when using current multi-planar anti-aliasing technique, a depth anti-aliasing algorithm based on classified mapping is proposed. As the imaging of solid-state volumetric true-3D display is characteristic of three-dimensional surface, two-dimensional curves generating from interception of YOZ (or XOZ) planar and three-dimensional surface are analyzed. A method of voxel decomposition based on adjacent multi-layers mapping is introduced, which conforms with human stereoscopic vision, as well as a type judgment algorithm of voxel mapping. The off-axis viewing angles under the conditions of depth continuity are expanded to some extent by means of self-adapted classified mapping of raw voxels. A solid-state volumetric true-3D display system is developed, and 3D images are tested on the system. The result shows that most 3D images have completely smooth surfaces out to large off-axis viewing angles (about 45° ), and a good depth continuity can be obtained.