High Dynamic Range Image Synthesis Techniques Based On A Camera Array

Compared with ordinary images,High Dynamic Range(HDR)images can show the details of real scene with a wide dynamic range on a typical display device.In HDR images,the details of high-brightness and low-brightness regions can be well preserved,which is closer to the reality,thus,they attract the attention of many scholars.In view of the current market price of equipment and technology maturity,hardware devices which can only capture images with low dynamic range will continue to occupy the consumer market in the next few decades.Thus,synthesizing a HDR image by using multi-exposure images is still viewed as the main direction.The main idea of the method is that different exposure images can reflect the detail information of different regions of the same scene,and the information provided by them can be supplemented by the synthesis,and an image can be obtained which keeps the intact brightness information of all regions.However,the traditional high dynamic range image synthesis techniques require the scene and light are both static.Where the event of dynamic scenes and dynamic lighting conditions happens,there will not be able to synthetize high dynamic range images normally.In addition,the traditional high dynamic range images are usually synthesized by an ordinary camera capturing the same scene though setting different exposure time.During the period,the camera is prone to jitter leading to blurring,and moving objects in the scene will cause ghost shadow.In order to solve these problems,this thesis studies the techniques of high dynamic range image synthesis based on a camera array,including image deblurring,multi-exposure image registration,high dynamic range image synthesis and image tone mapping.