作者: Thomas F. Schatzki , Pamela M. Keagy
DOI: 10.1117/12.25089
关键词: Pixel 、 Pixel intensity 、 Pattern recognition 、 Computer vision 、 Kernel (image processing) 、 Image resolution 、 Artificial intelligence 、 Biology
摘要: Radiograms of individual wheat kernels some which contained immature insects (weevils moths or borers) were captured using microscopy into a 512x512x8 frame buffer 16. 4 pm/pixel. The resulting image was reduced repeatedly to half-size produce collages containing 16 64 and 256 images. Half the one half no insect. Subjects shown sets 500 images on 13 " monitor asked determine whether an insect present. Since hidden are typically characterized by light inside dark ring (the hole) kernel we measured size recognizable region as length (major diameter) width hole in pixels image. Contrast represented square root variance pixel intensity along width. We find that recognition can be predicted 80-90 alone contrast add little. for 50 amounts 12. 20. 6 22. 10. 7 19. 80. weevils borers resp. It is suggested depends recognizing yet unknown features lost when beyond this size. 1."