Visual Measurement Method And Error Analysis Of Light-Field Camera

Light-field camera is a new imaging system based on the technology of computational photography.It can record the location and direction information of light in space simultaneously through one exposure.Based on the characteristics of multiple perspectives and refocusing,it can be used for visual measurement.However,in the process of visual measurement,the light field image is influenced by the specular reflection,as well as the influence of structural parameter error and random error,the error of visual measurement is increased.Therefore,we mainly focuses on the visual measurement methods and influence factors of error analysis of light-field camera.The main work of this thesis is listed as follows:(1)We systematically analyze the influence factors and measurement ranges of the EPI vision measurement,refocusing method and multi-view measurement method of light field.And the relationship between the visual measurement error and the actual distance,the length of the baseline,the focal length of the main lens as well as the high-light components are verified by experiments.(2)We propose a highlight removal algorithm of light-field aperture image,firstly,it extracts and matches the feature points of the multi-view image of light field by SIFT operator,then the high-light region is detected and removed by using the MSF image.The experimental results show that the proposing algorithm has better highlight removal effect and noise resistance,comparing with the existing algorithm.(3)We propose a visual measurement method based on multi-view combination of light field.We use the highlight removal algorithm to preprocess light field aperture images,and reduce the measurement error by the combination of multi-view.Comparing with the existing measurement methods,the experimental results show that our method has higher accuracy under the fixed structure of the micro lens array.