Estimation of efficiency of measurement of digital camera photosensor noise by automatic segmentation of non-uniform target method and the standard EMVA 1288

Abstract

In this paper important task of estimation of digital camera’s noise parameters is considered. Relation of accuracy of data obtained with digital camera and photosensor noise is discussed. Both standard European machine vision association EMVA 1288 and fast automatic segmentation of non-uniform target (ASNT) noise estimation methods are compared. Noise characteristics of machine vision PixeLink PL-B781F, scientific Retiga R6 and amateur mirrorless Canon EOS M100 cameras have been investigated. Accuracy of measurements, speed of calculation and experimental realization has been analyzed. Accuracy of temporal noise estimation by modified ASNT method is no less than that one for standard EMVA 1288. But the ASNT method can be implemented much faster than the standard EMVA 1288 even with additional frames for accuracy improvement.