Three problems in digital photography: Image sharpness, image interpolation, and image restoration.

摘要

We present three research topics related to digital photography:image sharpness, image interpolation, and image restoration.;In Chapter 1, we propose three device-independent reference-free sharpness metrics to measure the perceived sharpness of a digital image: ''Laplacian of Gaussian Contrast (LoGC)'', ''Average Edge Transition Slope (AETS)'', and ''Average Edge Transition Width (AETW)''. Results from psychophysical experiments show that our proposed metrics agree well with perceived sharpness. In order to compute the AETW and AETS, we develop an algorithm that can accurately extract edge normal profiles from any complex images. We also design and perform psychophysical tests to study sharpness detection threshold as well as sharpness preference. Our major conclusions are: 1) the sharpness detection threshold is relatively consistent across image contents, while the sharpness preference strongly depends on the image content; 2) the sharpness preference is consistently higher than the detection threshold across image contents, which implies that the average observer prefers a sharpened image to the original image.;In Chapter 2, we set out to improve an existing image interpolation algorithm--the Resolution Synthsis (ResSynth) algorithm. ResSynth interpolates sharper images than do the commonly used Bilinear and Bicubic interpolation; and it is computationally efficient. However, it has some deficiencies: 1) aggravated noise and JPEG artifacts; 2) halos; 3) occasional pixel errors around the edges. To overcome these problems, we modify ResSynth with three major procedures. We demonstrate that our New ResSynth algorithm significantly improves the image quality over ResSynth for a wide range of images.;In Chapter 3, we present an adaptive bilateral filter (ABF) for sharpness enhancement and noise removal. ABF sharpens an image by increasing the slope of the edges without producing overshoot or undershoot. Our new approach to slope restoration significantly differs from the previous slope restoration algorithms in that ABF does not involve detecting edges. Compared with the bilateral filter, ABF restored images are significantly sharper. Compared with an unsharp mask (USM) based sharpening method --- the Optimal USM (OUM), ABF restored edges are as sharp as those rendered by the OUM, but without halo. ABF also outperforms the bilateral filter and the OUM in noise removal.