Research On Image Computing Refocusing Method Based On Camera Array

The light field imaging developed in recent years can overcome the limitations of traditional imaging and exhibit excellent imaging performance.Compared with other light field imaging systems,camera arrays can increase the ability to acquire light fields by adding imaging units,and are expected to break the limitations of traditional imaging in terms of cost,volume and performance.Therefore,camera arrays have a good prospect.Refocusing is a basic function of the light field imaging system.By refocusing,the acquired light field image can be focused to any depth and the signal-to-noise ratio of the focused image is improved.This paper takes the array camera as the research background,briefly introduces the relevant knowledge of camera calibration and calibrates the scanning light field imaging platform built by the laboratory.It focuses on the refocusing technology based on camera array and has achieved certain research results.There are mainly the following aspects in this paper:It is analyzed how to correct a single camera scanning light field imaging system.When using calibration method to calibrate a single-camera scanning camera array,if only the translation stage is considered in the calibration process,the situation is a degraded model of calibration,and the corrected result is inaccurate.Therefore,it is necessary to add a checkerboard image of other angles as an aid in the correction to make the correction result substantially accurate.Using this method,the scanning platform developed by the laboratory is calibrated,the parameters of the camera at different positions are obtained,and the parameters of the camera can be used for subsequent refocusing process.The refocusing technology based on camera array is studied.Image degradation caused by interpolation is the main source of refocusing errors.Multiple interpolation results in a drop in image quality,and reduceing the number of interpolations can reduce the image degradation.Based on this consideration,a refocusing method based on re-projection is proposed and the influence of interpolation on image quality is analyzed.The experimental results of refocusing show that compared with the current homography-based focusing method,the proposed method does improve the quality of the focused image.The technique of focusing on an arbitrary surface based on camera array is further studied.Existing refocusing methods can only focus the acquired light field image on a plane,and in practice it is sometimes necessary to focus on the surface.Therefore,based on the refocusing method based on re-projection of one-time interpolation,a method of focusing a light field image on an arbitrary surface is proposed.First,we discussed how to focus the light field image on the tilted planes and analyze how to change the parameters to achieve the purpose of focusing on different tilted planes.Then we discussed the case of focusing on the surface.Taking the spherical and cylindrical surfaces as examples,we analyze how to determine the parameters of the spherical and cylindrical surfaces.Finally,the algorithm is verified by the real scene data obtained by the laboratory equipment.The results show that the image can be focused on different surfaces.