Quality Metrics of Digitally Derived Imagery and Their Relation to Interpreter Performance. III. Subjective Scaling of Hard-Copy Digital Imagery

摘要

Hard-copy digital imagery was studied with respect to subjective image quality. Trained Air Force photointerpreters judged the interpretability of 250 military scenes. The scenes varied in noise, blur, and scene content. The results showed that noise, blur, and scene content produce differential perceptions of interpretability. Many interactions were significant. Other analyses performed on the data showed that at least 62 categories should be used to scale interpretability, the correlation between information extraction performance and scale values for digital imagery is high, and multidimensional scaling can be used with limited utility in studying image quality. In general, from a review of the literature, digital imagery did not appear to be greatly different from standard analog imagery in terms of subjective quality or interpretability. (Author)