Digital document segmentation method

摘要

The method involves performing decomposition (E201) of wavelet pixels so as to obtain coefficients of detailed images, and partitioning (E202) a wavelet decomposition in a set of coefficient blocks according to a grid. A set of pixels of a document corresponding to the blocks is identified (E220). A histogram of pixel values is determined (E221). A determination is made (E232) if an area having pixels corresponding to the identified block is a text box by comparing data obtained from the histogram with two Gaussian threshold values. An independent claim is also included for a device for segmenting a digital document.