Research On Depth Acquisition And Application Based On Light Field Camera

With the development of light field imaging technology and the theory of light field,The light field camera is widely used in various fields.In the respect of depth acquisition,light field camera can achieve focusing after getting the light field images and obtain the multi-view images,which produces a variety of depth generation methods that are unique to itself,however some obvious shortcomings in quality and application of images of light field are still exposed.Thus,in this paper,we focus our research on improvements of depth acquistion and its application on digital refocusing by utilizing the light field camera.The innovation points are as following:1)An algorithm of digital refocusing of the digital Lytro light field camera is realized.Firstly,peak detection method is implemented to the high-exposure images to calibrate the center of micro lens array;Then the original light field image is remapped to get 4D light field data using the center coordinates data obtained in calibration;Finally,4D light field data re-integration and transformation are conducted to get the refocused images.The effectiveness of the proposed method is verified by the experiments.2)An optimized depth acquisition based on interactive image segmentation for is proposed.Firstly,an interactive method is used to segment the complex parts of the scene.And then,depth range of original scene is redefined by utilizing the segmented scene combined with defocusing depth image.Finally,remap the acquired depth map to the quantization level of 256 by combining defocus and correspondence.The depth range here is actually the refocus range of light field images.Using this method,the overall quality of the depth maps is improved without increasing the time of depth acquisition,which is conductive to post processing and application of depth images.The experiments indicate proposed algorithm can improve visual effect and objective quality.3)A method is proposed to combine defocusing images and depth maps to obtain the all-in-focus image.In this paper,original light field images are firstly obtained to acquire defocusing images in different depth after beingrefocused.Secondly,a simple depth map is generated by defocusing.Finally,the depth map is used to ensure the specific focus of every defocusing image and the all-in-focus image is ultimately received.This method avoids the traditional limitation of fusion algorithm and locates the exact focus region of defocusing images to generate all-in-focus images.The experiments indicate that the proposed algorithm can improve visual effect and objective quality.