Method and Good Estimators for Projection Uniformity Measurement

Francois Helt

An improved method for screen luminance uniformity measurement was proposed in a previous paper using a calibrated digital camera to capture a complete view of the screen in one shot. — As large screen projection exhibits vignetting, a model of the vignetting phenomenon is used in this paper to characterize the distribution of luminance values over the screen. The specific geometry of the projection is taken into account in the model by integrating the throw ratio as a parameter of the model. — The results give more detailed information and can be used to store and track the history of measures of multiple screens. — A similar approach has been explained for flat panel uniformity measurements for the broadcast industry. Once the vignetting of the given projection is compensated for, the process of computing a quality estimator is equivalent for both display systems. — Traditional point-wise measurements can be computed from both methods, but it is also worth trying to give more informative measurements and associated tolerances to obtain a useful measurement system. This was left open at the end of the last paper; we will explain how quality measures can be computed and justified. — The above mentioned broadcast paper proposes a quality estimator called the “uniformity score”. It is computed from the statistics of small patches evenly distributed all over the screen. It is based on standard deviation and scaled; questions arise about the scaling. — Deviations from a perfectly uniform projection may have multiple causes. Any uniformity measure must somehow be related to human perception of uniformity. Random deviations are less visible than structured luminance patterns. It is then logical to complement the standard deviation measure by a statistical measure which may indicate a biased error distribution. One such measure is proposed.

Published
2010-10
Content type
Original Research
Keywords
Uniformity, screen's measurement, vignetting
DOI
10.5594/M001398
ISBN
978-1-61482-944-7