Light field refocusing analysis of different interpolation algorithms

The light field data refocusing includes spatial domain refocusing and frequency domain refocusing. In the process of the refocusing, the interpolation calculation is needed, the refocusing precision is closely related to the interpolation precision, and the computational efficiency can be affected by the complexity of the interpolation algorithm. In the actual large amount of the image processing, the method with the highest computational efficiency is selected under the premise that the refocusing effect satisfies the accuracy requirement, and wherein the interpolation accuracy has a much greater influence on the frequency domain than the spatial domain. The principle of the spatial domain refocusing and the frequency domain refocusing, as well as the traditional three interpolation methods and the sinc function interpolation method were introduced. Different interpolation methods were used to experiment this two refocusing methods to compare their refocusing effect and calculate their efficiency. The experimental results show that in order to meet the requirement of the computational efficiency in actual applications, the linear interpolation method is the best method to optimize the spatial domain refocusing in the case of meeting the accuracy requirement, and the sinc interpolation with the sampling radius of 2 is optimal in the frequency domain. When the same image is refocused for multiple times, the frequency domain method is superior to the spatial domain method.