传统成像获取信息不足, 成像质量有一定局限性。为此, 提出了一种深度成像模型。模型包含深度矩阵、分解函数、散焦算子、自适应正则项等部分。深度矩阵的获取有双目立体视觉、结构光或飞行时间法等实现方法; 分解函数用于将图像按深度值的不同分割为若干子图像; 散焦算子可以通过深度散焦法来计算; 自适应正则项的引入能减少图像的阶梯效应, 增强图像的光滑性。通过局部标准差和局部平均梯度这两个评价指标检验深度成像模型的效果。实验结果表明, 深度成像模型效果显著。

Traditional imaging methods are subject to limitations on information acquisition, which brings about the deficiency of image quality. Therefore, a depth photography model was proposed here. This model contains a depth matrix, a decomposition function, a defocus operator, and an adaptive regularization term. The depth matrix could be estimated using the binocular stereo vision method, the structured light approach, or the time-of-flight algorithm. The decomposition function was used to segment the image into pieces according to the distinct depth values. The defocus operator was calculated through a defocus-from-depth method. The adaptive regularization term reduces the staircase effect and enhances image smoothness. Local standard deviations and local average gradients were used to evaluate the effectiveness of depth photography model. Experimental results demonstrate that the proposed model is effective.