Search

This paper presents a model that is then simplified to explain the temperature dependence of fixed pattern noise (FPN) in logarithmic complementary metal-oxide semiconductor (CMOS) image sensors. The simplified model uses the average dark response of pixels, which depends only on temperature, to help predict the FPN in the light response, which depends on temperature and illuminance. To calibrate a logarithmic camera, one requires images that are taken at different temperatures and illuminances, which need not be measured, of a uniform stimulus. To correct the FPN in an arbitrary image, one uses the simplified model parameters, which are estimated once by the calibration, and the average dark response, which is infrequently determined by closing the aperture. Through simulation (using mismatch data from a real CMOS process) and experiment (using a commercial logarithmic camera), an improvement is shown in the residual error per image, after calibration, when the proposed method is compared with a related method in the literature that does not account for temperature dependence. [ABSTRACT FROM AUTHOR]

Wide dynamic range logarithmic imagers can render naturally illuminated scenes while preserving detail and contrast information at a lower cost than high dynamic range linear sensors. However, the quality of the output is severely degraded by fixed pattern noise (FPN). Although previous FPN correction techniques can eliminate the dominant additive form of this noise, the contrast threshold of the imager over a wide illumination range is poor compared to the human visual system. In this paper, it is shown that a four-parameter model fits the measured characteristic response of wide dynamic range pixels over 11 decades of input current. A comparison of the responses of 200 pixels shows that there are significant variations in all four parameters. A procedure is described that allows the four pixel parameters to be obtained from the response of each pixel to five input currents. However, a much simpler procedure is shown to correct FPN, leading to a contrast threshold comparable to the human visual system over the five decades required to image wide-dynamic-range, naturally illuminated scenes. [ABSTRACT FROM AUTHOR]

Logarithmic CMOS image sensors encode a high dynamic range scene in a manner that roughly approximates human perception whereas linear sensors with equivalent quantization suffer from saturation or loss of detail. Moreover, the continuous response of logarithmic pixels permit high frame rates and random access, features that are useful in motion detection. This paper describes how to model, calibrate, and render pixel responses from a color logarithmic sensor into a standard color space. The work unifies color theory in conventional linear sensors and fixed pattern noise theory in monochromatic logarithmic sensors. Experiments with a Fuga 15RGB sensor demonstrate calibration and rendition using a Macbeth chart and neutral density filters. Color rendition of the sensor with an empirical model, tested over three decades of dynamic range, competes with conventional digital cameras, tested over 1.5 decades. Photodiode leakage currents complicate modeling and calibration and degrade rendition in dim lighting. [ABSTRACT FROM AUTHOR]

Presents information on a study which described a nonlinear model for the calibration of complementary metal-oxide semiconductor (CMOS) image sensors. Modeling of CMOS image sensors; Experimental results; Conclusions.

Photometric stereo and depth-map estimation provide a way to construct a depth map from images of an object under one viewpoint but with varying illumination directions. While estimating surface normals using the Lambertian model of reflectance is well established, depth-map estimation is an ongoing field of research and dealing with image noise is an active topic. Using the zero-mean Gaussian model of image noise, this paper introduces a method for maximum likelihood depth-map estimation that accounts for the propagation of noise through all steps of the estimation process. Solving for maximum likelihood depth-map estimates involves an independent sequence of nonlinear regression estimates, one for each pixel, followed by a single large and sparse linear regression estimate. The linear system employs anisotropic weights, which arise naturally and differ in value to related work. The new depth-map estimation method remains efficient and fast, making it practical for realistic image sizes. Experiments using synthetic images demonstrate the method's ability to robustly estimate depth maps under the noise model. Practical benefits of the method on challenging imaging scenarios are illustrated by experiments using the Extended Yale Face Database B and an extensive data set of 500 reflected light microscopy image sequences. [ABSTRACT FROM PUBLISHER]