A method to measure the modulation transfer function of Bayer filter color camera

Abstract

With the development of optoelectronic technologies, color cameras have been widely exploited in space remote sensing, earth observations from space, environmental monitoring, urban construction, and many other fields. Currently, most commercial color cameras use a single charge coupled device (CCD) or complementary metal-oxide-semiconductor (CMOS) sensor that has a Bayer color filter array (CFA) on its pixel surface to obtain red (R), green (G), or blue (B) samples. As a way of evaluating imaging quality, modulation transfer function (MTF) can provide a comprehensive and objective metric for camera imaging performance. In the conventional knife-edge method for color camera MTF measurement, a linear uniform sampling of the edge spread function (ESF) must be completed before a fast Fourier transform (FFT) can be applied. As the sampling rate becomes large, the number of pixel points on the line which is parallel to the knife-edge become less. So taking average of the pixel points to obtain ESF can be strongly affected by the noise of sensor. Therefore it is necessary to balance the influences of sampling rate and sensor noise on the MTF measurement, and the recommended sampling rate is 4-6. When the tilt angle of knife-edge has an error, the non-uniform sampling ESF can be obtained by the slanted knife-edge method. This leads to a variation in the results of the camera MTF on a spatial frequency scale and early cut-off. The best MTF results of camera can be obtained by rotating knife-edge, calculating MTF power under different tilt angles of knife-edge, and finding the maximum MTF power. And we propose an algorithm for Bayer filter color camera MTF measurement. The algorithm processing includes extracting R, G, B colors of knife-edge images; projection; differential operation; Hanning window filtration; FFT; correction; weighting combination of R, G, B colors MTF; MTF power calculation; optimal tilt angle of knife-edge estimation. To verify the accuracy of the proposed method, the weighting response factors of R, G, B colors are calibrated and an experimental setup for color camera MTF measurement is established. The knife-edge target is rotated in angle steps of 0.02, and the MTF results are calculated under different tilt angles of knife-edge within0.1 surrounding the estimate position by the proposed algorithm. The maximum differences of MTF results between the proposed method and fringe target method are 0.061 (Nyquist frequency fc) and 0.043 (fc/2), respectively. The results show that by searching the optimal tilt angle of knife-edge, the effect of non-uniform sampling on MTF result of color camera can be eliminated. Compared with the conventional method, the proposed method is superior for the measurement of the super-sampled MTF of color camera. Meanwhile, this method can also be applied to MTF measurements of radiographic systems, such as X-ray imaging system and other systems.