单幅散焦图像深度计算方法

Calculation Method of Depth in Single Defocused Image

由单幅图像获取深度是单目视觉测量中的一个难题。根据散焦图像的散焦原理,得到了单幅散焦测距扩散系数。基于单幅图像中不同区域的图像相关性,提出了替代二次成像要求的约束条件,即采用图像检测方法检测两个被测区域,结合散焦图像的各向异性扩散模型及区域扩散系数,分别对两个区域图像进行扩散实现。利用最小能量泛函求解了各个被测区域的各向异性图像扩散方程,进而得到了区域物体深度。在精度方面,实验结果与利用两幅散焦图像获取深度的传统方法相同。这种方法在测距过程中无需对相机参数进行调整,提高了单目测距的可操作性。该方法是对现有视觉测量方法的有力补充,能够为视觉测量技术提供更加宽广的应用前景。

Using a single image to acquire the depth information of an object is a great challenge in monocular vision.According to the defocusing principle of a defocused image,a single defocusing diffusion coefficient is deduced.Then,in the light of the correlation of different areas in the single image, a constraint condition for replacing the secondary imaging is proposed.In the constraint condition,an image detection method is used to detect two areas to be measured and the diffusion of the images of two areas is implemented respectively by incorporating the anisotropic diffusion model and area diffusion coefficient of the defocused image.Finally,a minimum energy function is used to solve the anisotropic image diffusion equations for each measured area and hence the depth of the object in the area is obtained.In accuracy,the experimental result is similar to that of the traditional method in which two defocus images are used to obtain the depth.Since the camera parameters need not to be adjusted in the measuring process,the operability of monocular ranging is improved.This method is a powerful alternative to the existing vision measuring methods and will have broader application prospects.