Degree of quantization and spatial addressability tradeoffs in perceived quality of color images

The objective of this thesis research was to investigate the tradeoffs between the number of quantization levels and spatial addressability of printed color images. Image quantization was done by employing the error-diffusion algorithm. The diffusion of error was performed in CMYK color space. The resulting images were printed on a color output device simulating different spatial addressabilities. To evaluate the perceived image quality, a psychophysical experiment was conducted followed by a statistical analysis of the experimental data. Based on the results of this analysis, the conclusions on the tradeoffs between the number of quantization levels and spatial addressability were drawn. It was determined that the tradeoffs were scene dependent with photographic scenes being able to sustain greater reduction in addressability without perceived image quality being decreased than graphics. The experiment showed that photographic scenes were sufficient to be printed with 5 bits per pixel per color at 100 dots per inch, and graphics with 3 bits per pixel per color at 300 dots per inch. If a single bits per color / dots per inch combination is to be named as the optimum combination equivalent to the best possible image for the given system (8bpc/300dpi), it would have to be 3bpc/300dpi. This combination was found to be equivalent to the quality of the best possible image at the normal viewing distance for all scenes in the experiment