基于双目相机的光栅立体印刷图像合成

目的研究一种基于双目相机的光栅立体图像合成方法。方法首先用双目相机采集场景中2个观察角度的二维平面图像数据。基于双目立体视觉理论,采用一种鲁棒性较强的基于区域分割的图像匹配方法,得到精度较高的深度图。然后分析序列视差图像的成像模型,建立一种基于双目图像对生成序列视差图像的方法,得到连续角度等间隔的序列图像。最后基于柱镜光栅的光学特性形成的立体印刷图像编码规则,对序列视差图像进行纵向条纹抽样分割,等间隔的抽取每幅序列视差图像中的对应列实现光栅立体图像的合成。结果验证了该光栅立体图像合成方法的有效性。结论基于双目相机的光栅立体合成方法,可以使立体印刷产品实现个性化、便捷化的即时输出。

Synthesis of Lenticular Printing Image Based on Binocular Camera

Objective To research a synthesis technology of Lenticular Printing Image based on binocular stereo camera. Methods Firstly, two-dimensional image data was collected from different viewing angles based on binocular camera. By studying the binocular stereo vision theory, we used a robust region stereo matching method based on color segmentation and got high-precision depth maps. By analyzing the imaging model of sequence parallax images, we established a new method of generating sequence parallax images based on binocular pairs and got sequence images with continuous angle and equal interval. At last, a suitable synthetic method based on optical properties of lenticular was used. The longitudinal stripes were splat for each of the sequence parallax images and then stripes from different sequence parallax composed the lenticular printing image. Results The effectiveness of the proposed method was verified by experiment. Conclusion The synthesis of lenticular printing image based on binocular stereo camera could make the personalized, convenient and real-time output three-dimensional printing products possible.