为了提高聚焦形貌恢复技术的重建精度,寻找最优的聚焦形貌恢复算法,提出一种图像离焦仿真技术,用于对聚焦形貌恢复算法的精度进行评估。绘制一个三维模型,并选择一幅纹理图,根据图像分辨率对模型表面进行等间距点采样,生成有序点云数据。对通过点采样获取的数据进行处理并映射纹理,生成一组理想的或者包含噪声的序列仿真图像。利用得到的序列图像对聚焦形貌恢复算法进行实验验证,将生成的深度数据与点采样得到的真实深度数据进行对比以准确地评价算法的质量。研究结果表明,图像离焦仿真技术能够有效评价聚焦形貌恢复算法的质量,并有助于寻找更稳定、精度更高的算法。

To improve the reconstruction accuracy of the focused morphology recovery technique and find the optimal focused morphology recovery algorithm, a simulation technology of image defocus is proposed in this paper, which is used to assess the reconstruction accuracy of the focused morphology recovery algorithm. A three-dimensional model is conducted and a texture image is selected. Then, the uniformly-spaced point sampling on the model surface according to the image resolution is performed and organized point clouds are generated. The acquired data obtained through point sampling is processed and the texture is mapped. At last, a set of sequence simulation images which contain the ideal image or images with noises are generated. The sequence images are used to verify the focused morphology recovery algorithm in experiments. The reconstructed depth data and the actual depth data are compared to accurately evaluate the performance of the proposed algorithm. The research results demonstrate that the image defocus simulation can assess the performance of the focused morphology recovery algorithm effectively and provide further support to find more algorithms with higher stability and accuracy.