Nathan Moroney

Color space selection for JPEG image compression

The Joint Photographic Experts Groups image compression algorithm has been shown to provide a very efficient and powerful method of compressing images. However, there is little substantive information about which color space should be utilized when implementing the JPEG algorithm. Currently, the JPEG algorithm is set up for use with any three-component color space. The objective of this research is to determine whether or not the color space selected will significantly improve the image compression. The RGB, XYZ, YIQ, CIELAB, CIELUV, and CIELAB LCh color spaces were examined and compared. Both numerical measures and psycho-physical ntechniques were used to assess the results. The final resuLts indicate that the device space, RGB, is the worst color space to compress images. In comparison, the nonlinear transforms of the device space, CIELAB and CIELUV, are the best color spaces to compress images. The XYZ, YIQ, and CIELAB LCh color spaces resulted in intermediate levels of compression.

An appearance uniformity metric for 3D printing

A method is presented for perceptually characterizing appearance non-uniformities that result from 3D printing. In contrast to physical measurements, the model is designed to take into account the human visual system and variations in observer conditions such as lighting, point of view, and shape. Additionally, it is capable of handling spatial reflectance variations over a materials surface. Motivated by Schrödingers line element approach to studying color differences, an image-based psychophysical experiment that explores paths between materials in appearance space is conducted. The line element concept is extended from color to spatially-varying appearances-including color, roughness and gloss-which enables the measurement of fine differences between appearances along a path. We define two path functions, one interpolating reflectance parameters and the other interpolating the final imagery. An image-based uniformity model is developed, applying a trained neural network to color differences calculated from rendered images of the printed non-uniformities. The final model is shown to perform better than commonly used image comparison algorithms, including spatial pattern classes that were not used in training.