Over the past two decades, there have been various studies on the distributions of the DCT coefficients for images. However, they have concentrated only on fitting the empirical data from some standard pictures with a variety of well-known statistical distributions, and then comparing their goodness of fit. The Laplacian distribution is the dominant choice balancing simplicity of the model and fidelity to the empirical data. Yet, to the best of our knowledge, there has been no mathematical justification as to what gives rise to this distribution. We offer a rigorous mathematical analysis using a doubly stochastic model of the images, which not only provides the theoretical explanations necessary, but also leads to insights about various other observations from the literature. This model also allows us to investigate how certain changes in the image statistics could affect the DCT coefficient distributions.

A new Image Systems Engineering Program (ISEP) has recently been launched at Stanford University. The program includes more than a dozen faculty participants drawn from four departments. The planned stages of growth of the Program are described, with emphasis on the anticipated respective roles of the university and industry.

A neural network model is presented in which each neuron performs a threshold logic function. The model always converges to a stable state when operating in a serial mode and to a cycle of length at most 2 when operating in a fully parallel mode. This property is the basis for the potential applications of the model, such as associative memory devices and combinatorial optimization. The two convergence theorems (for serial and fully parallel modes of operation) are reviewed, and a general convergence theorem is presented that unifies the two known cases. New relations between the neural network model and the problem of finding a minimum cut in a graph are obtained.>

This paper applies the Gram-Charlier series method to the caculation of error probabilties in digital optical receivers. This method allows the calculation of "exact" error probabilities including the effects of avalanche noise, thermal noise, and arbitrary posidetection processing filter. The predictions of this method are compared with those of a simple Gaussian approximation and with the Chernoff bounds. Finally, the effects of modal noise are included in the theory, and some specific cases are explored numerically.