作者: Damon M. Chandler , Nathan L. Dykes , Sheila S. Hemami
DOI: 10.1117/12.595614
关键词: Contrast (vision) 、 Compression ratio 、 Lossless compression 、 Artificial intelligence 、 Masking (art) 、 Quantization (image processing) 、 Visual masking 、 JPEG 2000 、 Mathematics 、 Pattern recognition 、 Computer vision 、 Wavelet
摘要: A visually lossless compression algorithm for digitized radiographs, which predicts the maximum contrast that wavelet subband quantization distortions can exhibit in reconstructed image such are undetectable, is presented. Via a psychophysical experiment, thresholds were measured detection of 1.15-18.4 cycles/degree five radiographs; results indicate radiographs impose image- and frequency-selective effects on detection. presented individual images based model visual masking. When incorporated into JPEG-2000 applied to suite images, be compressed manner at an average ratio 6.25:1, with some requiring ratios as low 4:1 great 9:1. The proposed thus yields require minimum bit-rate indistinguishable from original images. primary utility its ability provide image-adaptive compression, thereby avoiding overly conservative or aggressive compression.