%0 Conference Paper %B 15th International Conference on Pattern Recognition, 2000. Proceedings %D 2000 %T The statistics of optical flow: implications for the process of correspondence in vision %A Fermüller, Cornelia %A Aloimonos, J. %K Bias %K Computer vision %K correlation %K correlation methods %K energy-based method %K flow estimation %K Frequency estimation %K gradient method %K gradient methods %K Image analysis %K Image motion analysis %K Image sequences %K least squares %K least squares approximations %K Motion estimation %K Nonlinear optics %K Optical feedback %K optical flow %K Optical harmonic generation %K Optical noise %K Statistics %K Visual perception %X This paper studies the three major categories of flow estimation methods: gradient-based, energy-based, and correlation methods; it analyzes different ways of compounding 1D motion estimates (image gradients, spatio-temporal frequency triplets, local correlation estimates) into 2D velocity estimates, including linear and nonlinear methods. Correcting for the bias would require knowledge of the noise parameters. In many situations, however, these are difficult to estimate accurately, as they change with the dynamic imagery in unpredictable and complex ways. Thus, the bias really is a problem inherent to optical flow estimation. We argue that the bias is also integral to the human visual system. It is the cause of the illusory perception of motion in the Ouchi pattern and also explains various psychophysical studies of the perception of moving plaids. Finally, the implication of the analysis is that flow or correspondence can be estimated very accurately only when feedback is utilized %B 15th International Conference on Pattern Recognition, 2000. Proceedings %I IEEE %V 1 %P 119-126 vol.1 - 119-126 vol.1 %8 2000/// %@ 0-7695-0750-6 %G eng %R 10.1109/ICPR.2000.905288 %0 Conference Paper %B Computer Vision and Pattern Recognition, 1999. IEEE Computer Society Conference on. %D 1999 %T Independent motion: the importance of history %A Pless, R. %A Brodsky, T. %A Aloimonos, J. %K aerial visual surveillance %K background image %K Fluid flow measurement %K Frequency measurement %K History %K Motion detection %K Motion estimation %K Motion measurement %K Noise measurement %K Optical computing %K Optical noise %K spatiotemporal image intensity gradient measurements %K Spatiotemporal phenomena %K Surveillance %K Video sequences %X We consider a problem central in aerial visual surveillance applications-detection and tracking of small, independently moving objects in long and noisy video sequences. We directly use spatiotemporal image intensity gradient measurements to compute an exact model of background motion. This allows the creation of accurate mosaics over many frames and the definition of a constraint violation function which acts as an indication of independent motion. A novel temporal integration method maintains confidence measures over long subsequences without computing the optic flow, requiring object models, or using a Kalman filler. The mosaic acts as a stable feature frame, allowing precise localization of the independently moving objects. We present a statistical analysis of the effects of image noise on the constraint violation measure and find a good match between the predicted probability distribution function and the measured sample frequencies in a test sequence %B Computer Vision and Pattern Recognition, 1999. IEEE Computer Society Conference on. %I IEEE %V 2 %P 97 Vol. 2 - 97 Vol. 2 %8 1999/// %@ 0-7695-0149-4 %G eng %R 10.1109/CVPR.1999.784614 %0 Conference Paper %B Computer Vision and Pattern Recognition, 1999. IEEE Computer Society Conference on. %D 1999 %T Statistical biases in optic flow %A Fermüller, Cornelia %A Pless, R. %A Aloimonos, J. %K Distributed computing %K Frequency domain analysis %K HUMANS %K image derivatives %K Image motion analysis %K Image sequences %K Least squares methods %K Motion estimation %K Optical computing %K Optical distortion %K optical flow %K Optical noise %K Ouchi illusion %K perception of motion %K Psychology %K Spatiotemporal phenomena %K statistical analysis %K systematic bias %K total least squares %X The computation of optical flow from image derivatives is biased in regions of non uniform gradient distributions. A least-squares or total least squares approach to computing optic flow from image derivatives even in regions of consistent flow can lead to a systematic bias dependent upon the direction of the optic flow, the distribution of the gradient directions, and the distribution of the image noise. The bias a consistent underestimation of length and a directional error. Similar results hold for various methods of computing optical flow in the spatiotemporal frequency domain. The predicted bias in the optical flow is consistent with psychophysical evidence of human judgment of the velocity of moving plaids, and provides an explanation of the Ouchi illusion. Correction of the bias requires accurate estimates of the noise distribution; the failure of the human visual system to make these corrections illustrates both the difficulty of the task and the feasibility of using this distorted optic flow or undistorted normal flow in tasks requiring higher lever processing %B Computer Vision and Pattern Recognition, 1999. IEEE Computer Society Conference on. %I IEEE %V 1 %P 566 Vol. 1 - 566 Vol. 1 %8 1999/// %@ 0-7695-0149-4 %G eng %R 10.1109/CVPR.1999.786994