A rapid method is developed for two-dimensional smooth¬ing and edge-sharpening by the least-squares fitting of a function to a limited area of the data. This convolution or matrix weighting is applied at each point of the data set to yield a smoothed or a sharpened image. Weighting matrices for 3 x 3, 5 x 5, and 7x7 point fitting areas are provided for polynomial function fits of all degrees up to the high¬est degree determinable. For the 7x7 point fitting area weights for fitting functions of up to the quartic in both dimensions are supplied. Application of the 5x5 point quadratic fit smoothing to a nuclear medicine image is shown as an example.
展开▼
