A general technique for the recovery of signicant

image features is presented. The technique is based on

the mean shift algorithm, a simple nonparametric pro-

cedure for estimating density gradients. Drawbacks of

the current methods (including robust clustering) are

avoided. Feature space of any nature can be processed,

and as an example, color image segmentation is dis-

cussed. The segmentation is completely autonomous,

only its class is chosen by the user. Thus, the same

program can produce a high quality edge image, or pro-

vide, by extracting all the signicant colors, a prepro-

cessor for content-based query systems. A 512  512

color image is analyzed in less than 10 seconds on a

standard workstation. Gray level images are handled

as color images having only the lightness coordinate-A general technique for the recovery of sig ni cannot image features is presented. The techni que is based on the mean shift algorithm, a simple nonparametric pro-cedure for estimat ing density gradients. Drawbacks of the curren t methods (including robust clustering) are av oided. Feature space of any nature can be proces sed, and as an example, color image segmentation is dis-cussed. The se gmentation is completely autonomous. only its class is chosen by the user. Thus, the same program can produce a high quality edge image, or pro-vide. by extracting all the signi cannot colors, a prepro- cessor for content-based query syste ms. A 512,512 color image is analyzed in less than 10 seconds on a standard workstation. Gray 4ISR l images are handled as color images having only the lightness c