Dark-frame subtraction

About: Dark-frame subtraction is a research topic. Over the lifetime, 1216 publications have been published within this topic receiving 20763 citations.

Sensor noise measurement in the presence of a flickering illumination

Abstract: Raw data from a digital imaging sensor are impaired by a heteroscedastic noise, the variance of pixel intensity linearly depending on the expected value. The most natural way of estimating the variance and the expected value at a given pixel is certainly empirical estimation from the variations along a stack of images of any static scene acquired at different times under the same camera setting. However, the relation found between the sample variance and the sample expectation is actually not linear, especially in the presence of a flickering illumination. The contribution of this paper is twofold. First, a theoretical model of this phenomenon shows that the linear relation changes into a quadratic one. Second, an algorithm is designed, which not only gives the parameters of the expected linear relation, but also the whole set of parameters governing an image formation, namely the gain, the offset and the readout noise. The rolling shutter effect is also considered.

Arbitrary background picture segmentation

Abstract: In this paper, we present a new method for real time arbitrary background picture segmentation. We consider the following picture conditions: color picture, noisy picture, scene light changes, and still image arbitrary background. They are typical for many applications, e.g. for video security system, videophone, videoconference, V-commerce, etc. A set-theoretic approach has been used for picture model creation, adaptive picture processing and noise reduction. The presented method main problem, namely noise immunity and resolving power, is investigated by using picture clustering. The both luminance and chrominance picture components use allows to avoid an influence of "undesirable" object light changes and noise. It is shown that for arbitrary background picture segmentation, the required signal-to-noise ratio must be from 40 dB to 44 dB, depending on the frames change rate.

Methods for eliminating artifacts in two-dimensional optical metrology

Abstract: Methods for eliminating artifacts in two-dimensional optical metrology utilizing the interline CCD detectors are based on a dark-subtraction principle. The self-dark subtraction method takes advantage of strong correlation between the noise patterns in illuminated and dark regions within the same image. Image artifacts are removed and the S/N ratio is improved significantly by subtraction of selected dark region of the image from the illuminated one within the same frame. The dark-frame subtraction technique reduces a “smear” effect by applying a digital processing based on subtraction of the dark frame images from the normal light frame images. A combination of these methods significantly improves performance of two-dimensional optical metrology systems such as spectrometers, ellipsometers, beam profile reflectometers/ellipsometers, scatterometers and spectroscopic scatterometers.

A new hybrid filtering technique for removal of impulse noise from digital images

Abstract: Removing impulse noise from digital image is a very active research area in digital image processing. In recent years, technological development has significantly improved in analyzing digital images. This paper proposes a new hybrid filtering technique (HMMF) for the removal of impulse noise from digital images, by topological approach. The quality of the noise reduction in images is measured by the statistical quantity measures: Root Mean Square Error (RMSE) and Peak Signal-to-Noise Ratio (PSNR). The performance of this filter on images tainted with various noise levels of impulse noise is compared with some existing filtering techniques.

Denoising photographs using dark frames optimized by quadratic programming

Abstract: Photographs taken with long exposure or high ISO setting may contain substantial amounts of noise, drastically reducing the Signal-To-Noise Ratio (SNR). This paper presents a novel optimization approach for denoising. It is based on a library of dark frames previously taken under varying conditions of temperature, ISO setting and exposure time, and a quality measure or prior for the class of images to denoise. The method automatically computes a synthetic dark frame that, when subtracted from an image, optimizes the quality measure. For specific choices of the quality measure, the denoising problem reduces to a quadratic programming (QP) problem that can be solved efficiently. We show experimentally that it is sufficient to consider a limited subsample of pixels when evaluating the quality measure in the optimization, in which case the complexity of the procedure does not depend on the size of the images but only on the number of dark frames. We provide quantitative experimental results showing that our method automatically computes dark frames that are competitive with those taken under idealized conditions (controlled temperature, ISO setting, exposure time, and averaging of multiple exposures). We provide application examples in astronomical image denoising. The method is validated on two CMOS SLRs.