Transformation between continuous and discrete representations of images: a perceptual approach.

The transformation of image signals between the continuous domain of the real world and the discrete domain of modern data processing can have a significant effect on quality and efficiency. Analysis based on perceptual rather than mathematical considerations has been carried out. A series of experiments based on the analysis has shown that substantial improvement over usual techniques can be achieved by the use of a cascade of a presharpening filter combined with Gaussian presampling and interpolation filters. The resulting ``sharpened Gaussian'' filter, although not exactly circularly symmetrical, gives a high degree of isotropy. Each element in the cascade is separable, so that computational efficiency is high. A favorable tradeoff is achieved among sharpness, smoothness, and the effects of aliasing. Subjective testing, in comparison with other commonly used filters, has shown the clear superiority of this filter.

Comparison of interpolating methods for image resampling.

When resampling an image to a new set of coordinates (for example, when rotating an image), there is often a noticeable loss in image quality. To preserve image quality, the interpolating function used for the resampling should be an ideal low-pass filter. To determine which limited extent convolving functions would provide the best interpolation, five functions were compared: A) nearest neighbor, B) linear, C) cubic B-spline, D) high-resolution cubic spline with edge enhancement (a = -1), and E) high-resolution cubic spline (a = -0.5). The functions which extend over four picture elements (C, D, E) were shown to have a better frequency response than those which extend over one (A) or two (B) pixels. The nearest neighbor function shifted the image up to one-half a pixel. Linear and cubic B-spline interpolation tended to smooth the image. The best response was obtained with the high-resolution cubic spline functions. The location of the resampled points with respect to the initial coordinate system has a dramatic effect on the response of the sampled interpolating function the data are exactly reproduced when the points are aligned, and the response has the most smoothing when the resampled points are equidistant from the original coordinate points. Thus, at the expense of some increase in computing time, image quality can be improved by resampled using the high-resolution cubic spline function as compared to the nearest neighbor, linear, or cubic B-spline functions.

Scaling binary images with the telescoping template.

The importance of enlarging and reducing two-level images such as graphical and documentary matter by digital means continues to grow as more such images are digitally represented. A nonlinear scaling scheme is devised which exploits the simplicity of this binary nature, treating images logically instead of arithmetically; a convolution-like effect is achieved without a single addition or multiplication! This method yields high-fidelity digital scaling and meets the objectives of being fast, conducive to hardware realization, and void of special pre-encoding requirements.

An interactive image processing system.

A multiuser multiprocessing image processing system has been developed. It is an interactive picture manipulation and enhancement facility which is capable of executing a variety of image processing operations while simultaneously controlling real-time input and output of pictures. It was designed to provide a reliable picture processing system which would be cost-effective in the commercial production environment. Additional goals met by the system include flexibility and ease of operation and modification.