A simple model of the phase-detection autofocus device based on the partially masked sensor pixels is described. The cross-correlation function of the half-images registered by the masked pixels is proposed as a focus...A simple model of the phase-detection autofocus device based on the partially masked sensor pixels is described. The cross-correlation function of the half-images registered by the masked pixels is proposed as a focus function. It is shown that—in such setting—focusing is equivalent to searching of the cross-correlation function maximum. Application of stochastic approximation algorithms to unimodal and non-unimodal focus functions is shortly discussed.展开更多
Efficiency of the autofocusing algorithm implementations based on various orthogonal transforms is examined. The algorithm uses the variance of an image acquired by a sensor as a focus function. To compute the estimat...Efficiency of the autofocusing algorithm implementations based on various orthogonal transforms is examined. The algorithm uses the variance of an image acquired by a sensor as a focus function. To compute the estimate of the variance we exploit the equivalence between that estimate and the image orthogonal expansion. Energy consumption of three implementations exploiting either of the following fast orthogonal transforms: the discrete cosine, the Walsh-Hadamard, and the Haar wavelet one, is evaluated and compared. Furthermore, it is conjectured that the computation precision can considerably be reduced if the image is heavily corrupted by the noise, and a simple problem of optimal word bit-length selection with respect to the signal variance is analyzed.展开更多
The problem of nonparametric identification of a multivariate nonlinearity in a D-input Hammer- stein system is examined. It is demonstrated that if the input measurements are structured, in the sense that there exist...The problem of nonparametric identification of a multivariate nonlinearity in a D-input Hammer- stein system is examined. It is demonstrated that if the input measurements are structured, in the sense that there exists some hidden relation between them, i.e. if they are distributed on some （unknown） d-dimensional space M in IRD, d 〈 D, then the system nonlinearity can be recovered at points on M with the convergence rate O（n-1/（2＋d）） dependent on d. This rate is thus faster than the generic rate O（n-1/（2＋D）） achieved by typical nonparametric algorithms and controlled solely by the number of inputs D.展开更多
基金supported by the NCN grant UMO-2011/01/B/ST7/00666.
文摘A simple model of the phase-detection autofocus device based on the partially masked sensor pixels is described. The cross-correlation function of the half-images registered by the masked pixels is proposed as a focus function. It is shown that—in such setting—focusing is equivalent to searching of the cross-correlation function maximum. Application of stochastic approximation algorithms to unimodal and non-unimodal focus functions is shortly discussed.
基金supported by the NCN grant UMO-2011/01/B/ST7/00666.
文摘Efficiency of the autofocusing algorithm implementations based on various orthogonal transforms is examined. The algorithm uses the variance of an image acquired by a sensor as a focus function. To compute the estimate of the variance we exploit the equivalence between that estimate and the image orthogonal expansion. Energy consumption of three implementations exploiting either of the following fast orthogonal transforms: the discrete cosine, the Walsh-Hadamard, and the Haar wavelet one, is evaluated and compared. Furthermore, it is conjectured that the computation precision can considerably be reduced if the image is heavily corrupted by the noise, and a simple problem of optimal word bit-length selection with respect to the signal variance is analyzed.
文摘The problem of nonparametric identification of a multivariate nonlinearity in a D-input Hammer- stein system is examined. It is demonstrated that if the input measurements are structured, in the sense that there exists some hidden relation between them, i.e. if they are distributed on some （unknown） d-dimensional space M in IRD, d 〈 D, then the system nonlinearity can be recovered at points on M with the convergence rate O（n-1/（2＋d）） dependent on d. This rate is thus faster than the generic rate O（n-1/（2＋D）） achieved by typical nonparametric algorithms and controlled solely by the number of inputs D.
