IMAGE ANALYSIS TO OBTAIN TYPEFACE INFORMATION

Dominant typeface, including point size, of text in an image is identified by measuring distances across connected components in each of plural directions to obtain distribution data. The distribution data indicate a distribution of distances across connected components in each direction for the image. The distribution data are then used to obtain typeface data indicating the dominant typeface of characters in the text. Distance can be measured from each of a set of pixels that are inside or on edges of connected components to an edge of a connected component. The distances for each direction can be combined, such as by obtaining an average or other measure of central value, so that the combined distances indicate the distribution of distances in the direction. The distribution data can be used to form a profile that can be compared or clustered with other profiles to identify similar profiles or to find a matching profile. Skew can be obtained and used in forming profiles that are independent of skew. The comparisons can be independent of differences in character scale. A machine can store a set of model profiles of known typeface and point size in memory and can compare a profile for an image with each model profile to find the closest match, providing an identifier of its dominant typeface and point size. A machine could also store an input image in memory, obtain data indicating its dominant typeface, modify the stored data according to the dominant typeface, and provide an output image based on the modification. Profiles can be stored or transmitted, and a product can include a model profile together with code for operating a processor to obtain a profile for an image and code for operating the processor to compare the profile for the image with the model profile.