摘要
为解决传统相机空间变化离焦去模糊的难题,提出了一种基于模糊映射图和L1-2优化的空间变化离焦去模糊方法。首先对相机的离焦模型进行研究与分析,概括了圆盘离焦模型和高斯离焦模型的原理、特性和适用范围;然后利用边缘属性对局部对比度先验条件进行修改,结合模糊边缘的梯度信息得到模糊映射图,并利用一种向导滤波的方法去除边缘附近的歧义性和映射图中的噪声,获得一幅更好的模糊映射图;最后采用一种合并了类Tikhonov规整化和TV规整化的L1-2优化对图像进行去模糊,采用变量分裂和惩罚的方法完成L1-2的优化,并利用尺度选择和图像重构方法得到全聚焦图像。基于人工合成图像和实际拍摄图像的实验结果表明,所提方法性能优于当前技术条件下的空间不变去模糊方法和空间变化去模糊方法。
To solve the spatially-varying out-of-focus deblurring problem of a conventional camera, a spatially- varying out-of-focus deblurring approach based on blur map and L1-2 optimization is proposed. Firstly, the defocusing model of camera is studied and analyzed, and the principle, characteristics and applicability of circular defocusing model and Gaussian defocusing model are presented. And then, the local contrast prior condition is modified using the edge properties, the blur map is obtained in combination with the gradient information of fuzzy edge, and a guided filtering method is adopted to remove the ambiguity around edges and the noise of blur map, thus obtain a better blur map. Finally, L1-2 optimization that combines Tikhonov-like regularization and TV regularization is applied to image deburring, which is accomplished with variable splitting and punishment method, and the all-in-focus image is obtained using scale selection and image reconstruction. The experimental results based on both synthetic image and real image show that the proposed approach outperforms the state-of-the-art space-invariant and spatially-varying deblurring methods.
出处
《电光与控制》
北大核心
2015年第9期91-95,共5页
Electronics Optics & Control
基金
国家自然科学基金(61175120)
关键词
景深
空间变化离焦去模糊
高斯离焦模型
模糊映射图
线扩展函数
depth of field
spatially-varying out-of-focus deblurring
Gaussian defocusing model
blur map
line spread function
