抄録
Graphic noise properties are usually evaluated using the Wiener spectrum. However, this approach is not appropriate for periodic noise for two reasons. One is that it takes infinite values at the spatial frequencies of periodic noise. The other is that when adopting a numerical integration, it allows unstable values at each spatial frequency, depending on the integral region. Introducing three types of spectra (W1, W2, and W3) in connection with the Wiener spectrum, we propose a practical approach to evaluation of periodic noise. Graphic images sometimes contain noise that is not totally random and not perfectly periodic. Therefore, using the Wiener spectrum and the W1 spectrum, we also propose two factors for evaluation of the degree of randomperiodicity of noise containing periodic signals.
