Visual Model of Modulation Transfer Function for Assimilation Phenomenon by Gaussian Filter

Abstract

This paper describes an assimilation phenomenon on sample images and proposes an analytical model by Gaussian filter for this phenomenon. First, the frequency spectrum of a sample pattern is formulated by the discrete Fourier transform. Next, human visual characteristics are formulated by Gaussian band pass filter. Thirdly, the concept of a noise level is introduced. Then, the two-dimensional model of the relations is discussed qualitatively and quantitatively. Finally, the verification experiments by subjective evaluation and measuring the visual distance was conducted. The results of the proposed model are very similar to the results of subjective evaluation under the reasonable conditions; an attenuation coefficient of 200 and a cutoff frequency of 5 cycles/degree in the visual BPF model, and a noise level of −40 dB.