Optimization of Luminance Computational Model of HDRI by the Multi-linear Regression Method

Abstract

The luminance measurement based on the High Dynamic Range Image (HDRI) is currently the mainstream luminance test scheme, and the research on the related luminance computational model has also attracted great attention. The effectiveness of this technique mainly depends on the luminance computational model, which is established by linear regression method or HDRscope software method currently. However, both methods have shortcomings, such as complex experiments and low accuracy. In order to address these problems, we creatively proposed the use of multi-linear regression method to optimize the previous HDRI luminance computational model. The linear regression method, the HDRscope software method and the multi-linear regression method are systematically compared in theory and experiment, which proves the applicability and effectiveness of multi-linear regression method. Compared to the traditional linear regression method, it reduces the dependence on the test equipment, simplifies the test process, and can ensure the accuracy of the test results effectively; compared to the HDRscope software method, it improves the accuracy rate significantly. Therefore, the utilization of multi-linear regression method to construct and optimize the HDRI luminance computational model can not only ensure the accuracy of the test results, simplify the process, but also reduce the dependence on the colorimeter. In practical applications, this will be an effective and reliable strategy.