Unevenness is an issue in high-resolution image sensors. The unevenness shows up as black and white bands extending horizontally on a white back-ground. The degree of unevenness is ranked by visual observation, but visual judgement is an ambiguous, time-consuming, and nerve-wracking Process. A new method was developed to solve these problems. The outputs from the pixels were expressed as two-dimensionally discrete orthogonal polynomials. The coefficients in the polynomials were converted to characteristic items deemed appropriate for visual judgment. The T method was used to relate the items to visual judgment. A new screening method called the two-item round-robin system was used to screen the characteristic items for better accuracy.