作者: Tingting Zhang , Ming Yu , Yingchun Guo , Yi Liu
DOI: 10.3390/INFO10030111
关键词: Image scaling 、 Pattern recognition 、 Structural similarity 、 Similarity (network science) 、 Image quality 、 Feature (computer vision) 、 Artificial intelligence 、 Image (mathematics) 、 Mean opinion score 、 Computer science 、 Correlation
摘要: In targeting the low correlation between existing image scaling quality assessment methods and subjective awareness, a content-aware retargeted algorithm is proposed, which based on structural similarity index. this paper, index, that is, local algorithm, can measure different sizes of same proposed. The Speed Up Robust Feature (SURF) used to extract content loss degree. significant area ratio calculated by extracting saliency region function obtained linear fusion. CUHK database MIT RetargetMe database, compared with four representative algorithms other latest kinds algorithms, experiment proves proposed has higher Mean Opinion Score (MOS) values corresponds result human assessment.