| Summary: | A monochromatic image can be represented by a real function f(x,y,t) of two spatial variables and time. Picture transmission research is directed at finding methods of transmitting a facsimile of f(x,y,t) at minimum cost which still satisfies some quality criteria. The ensemble of source, functions, and the measures of cost and quality are defined by the application. For many applications, the subjective quality is an important measure of performance.
Generally, it is difficult if not impossible to find a tractable analytic relationship between the source and system parameters, and the subjective quality. Thus,for picture processing research, equipment must be available for simulating proposed systems so that pictures can be produced and estimates made of the subjective quality.
This thesis will describe a system which was devised for studying low resolution picture transmission systems. The system employs an image dissector to input picture data to a PDP-9 computer. The data can be stored on magnetic tape, processed and then displayed on a picture output device which employs a Tektronix 561 Oscilloscope. A set of general purpose programs was developed for manipulating picture data using this hardware configuration.
Three test pictures were read in and particular programs were written to calculate some picture statistics. The statistics include signal histograms, auto-covariance functions and, some run length statistics for the two level material. These statistics were then used as design information for some simulation programs.
A program was written to simulate Pulse Code Modulation (PCM) and Differential Pulse Code Modulation (DPCM) systems in order to evaluate the subjective effect of channel errors on pictures transmitted by PCM and DPCM. An attempt was also made to develop a simple two-dimensional filter program to reduce the effect of channel noise. Since the channel noise in the PCM and DPCM systems produced two-dimensional noises which were very different in appearance, two filtering routines were necessary.
Subjective rating tests were conducted to evaluate the filtered and unfiltered PCM and DPCM pictures. For the unfiltered systems, these tests indicate that for good channels, the DPCM system is better than the PCM system while for very bad channels, the pictures produced by the PCM system are subjectively better than those of a DPCM system using the same number of bits. It was found also that the PCM filter proposed, improved the noisy pictures while the proposed DPCM filter produced no improvement and even lowers the quality of some of the pictures. === Applied Science, Faculty of === Electrical and Computer Engineering, Department of === Graduate
|
|---|
