Expectation-maximization algorithms for image processing using multiscale models and mean-field theory, with applications to laser radar range profiling and segmentation

We describe a new class of computationally efficient algorithms designed to solve incomplete-data problems frequently encountered in image processing and computer vision. The basis of this framework is the marriage of the exp ectation-maximization (EM) procedure with two powerful methodologies. In pa rticular, we have incorporated optimal multiscale estimators into the EM pr ocedure to compute estimates and error statistics efficiently. In addition, mean-field theory (MFT) from statistical mechanics is incorporated into th e EM procedure to help solve the computational problems that arise from our use of Markov random-field (MRF) modeling of the hidden data in the EM for mulation. We have applied this, algorithmic framework and shown that it is effective in solving a wide variety of image-processing and computer-vision problems. We demonstrate the application of our algorithmic framework to s olve the problem of simultaneous anomaly detection, segmentation, and objec t profile estimation for noisy and speckled laser radar range images. (C) 2 001 Society of Photo-Optical Instrumentation Engineers.