The hue saturation intensity (HSI) color model resembles the color sensing properties of human vision. Thus, it is more suitable to color image analysis than other color models. However, the hue (H) component is a periodic angular signal, to which the conventional signal processing algorithms cannot be applied without producing artifacts. We have developed an algorithm of generalized filtering operations on the angular signals, such as hue images. The developed method divides the input angular data into four groups according to the sign of filter coefficients and the range of angular signals. Each group is processed separately, and the results from the four groups are combined together. In order to verify our algorithm, we have applied it to the spatial smoothing and gradient processing of hue images, (C) 2001 Society of Photo Optical Instrumentation Engineers.