Mixed-type documents include text, drawings and graphics regions. It is obv
ious that a technique that can reduce the number of the gray-levels in acco
rdance with the type of each document region could be important for many do
cument applications, such as storage, transmission and recognition. To solv
e this problem, this paper proposes a new method, called the document multi
thresholding technique. The method is based on a page layout analysis (PLA)
technique and on a neural-network multilevel threshold-selection approach.
The proposed technique is applicable to any mixed-type document and achiev
es document multithresholding by taking advantage of the types of the docum
ent blocks. Thus, in the final document different block types are stored wi
th the appropriate and limited numbers of pray-level values. The proposed m
ethod includes two main steps. First, a PLA technique is applied, which cla
ssifies the document blocks into text, line-drawing and graphics regions. I
n the second stage, a new neural-network multithresholding technique is app
lied to each of the document blocks. In text and line-drawing blocks, only
one threshold is determined, whereas in the graphics blocks the optimal num
ber of thresholds is first determined. The performance of the method has be
en extensively tested on a variety of documents. Several examples illustrat
e the strength and the effectiveness of the proposed methodology. (C) 2000
Elsevier Science Ltd. All rights reserved.