Date of Award
Master of Science
Electrical and Computer Engineering
The purpose of image data compression is to represent data efficiently without loss of information. This involves identification and removal of unnecessary information. Uncompressed image data is typically represented in such a way so that it is highly redundant. Need for data reduction arises due to limitation on storage space or transmission time. Although the storage capacities of magnetic media increases, the demand for data compression has been growing steadily. The Nuclear Regulatory Commission requires that the radiographs be stored for 100 years. The film radiograph degrades due to aging. To avoid this generally the radiograph is digitized between 35 and 100 micron spatial resolution and 12 bits. For a 11x14 inch radiograph this requires on the order of 30 Mbytes for storage. Data compression is necessary to increase the number of images that can be stored. Various factors used in the evaluation of compression are the amount of compression provided, speed of compression and decompression, memory requirements and the mean square error (MSE). Since the radiographs are viewed by the human eye, it is very important that the compression does not introduce any artifacts that are visible. It is necessary to evaluate the visual impact of the error due to compression. In this thesis, a method is presented which calculates the visual distortion of the compressed image as compared to the original image. This method is based on a model of the human eye.
Digital Repository @ Iowa State University, http://lib.dr.iastate.edu
Ramaswami, Vivek, "Detectability model for the evaluation of lossy compression methods on radiographic images" (1996). Retrospective Theses and Dissertations. 250.