Seminar Topics & Project Ideas On Computer Science Electronics Electrical Mechanical Engineering Civil MBA Medicine Nursing Science Physics Mathematics Chemistry ppt pdf doc presentation downloads and Abstract

TEXT SEGMENTATION FOR MRC DOCUMENT COMPRESSION



ABSTRACT:
The mixed raster content (MRC) standard (ITU-T T.44) specifies a framework for document compression which can dramatically improve the compression/quality tradeoff as compared to traditional lossy image compression algorithms.
The key to MRC compression is the separation of the document into foreground and background layers, represented as a binary mask. Therefore, the resulting quality and compression ratio of a MRC document encoder is highly dependent upon the segmentation algorithm used to compute the binary mask.
In this paper, we propose a novel multiscale segmentation scheme for MRC document encoding based upon the sequential application of two algorithms. The first algorithm, cost optimized segmentation (COS), is a blockwise segmentation algorithm formulated in a global cost optimization framework.
The second algorithm, connected component classification (CCC), refines the initial segmentation by classifying feature vectors of connected components using an Markov random field (MRF) model. The combined COS/CCC segmentation algorithms are then incorporated into a multiscale framework in order to improve the segmentation accuracy of text with varying size.
In comparisons to state-of-the-art commercial MRC products and selected segmentation algorithms in the literature, we show that the new algorithm achieves greater accuracy of text detection but with a lower false detection rate of nontext features.
We also demonstrate that the proposed segmentation algorithm can improve the quality of decoded documents while simultaneously lowering the bit rate.

TEXT SEGMENTATION FOR MRC DOCUMENT COMPRESSION

[attachment=20388]

INTRODUCTION

WITH the wide use of networked equipment such as computers,
scanners, printers and copiers, it has become
more important to efficiently compress, store, and transfer large
document files. For example, a typical color document scanned
at 300 dpi requires approximately 24Mbytes of storage without
compression. While JPEG and JPEG2000 are frequently used
tools for natural image compression, they are not very effective
for the compression of raster scanned compound documents
which typically contain a combination of text, graphics, and natural
images. This is because the use of a fixed DCT or wavelet
transformation for all content typically results in severe ringing
distortion near edges and line-art.


COS
The COS algorithm is a block-based segmentation algorithm
formulated as a global cost optimization problem. The COS algorithm
is comprised of two components: blockwise segmentation
and global segmentation. The blockwise segmentation divides
the input image into overlapping blocks and produces an
initial segmentation for each block. The global segmentation is
then computed from the initial segmented blocks so as to minimize
a global cost function, which is carefully designed to favor
segmentations that capture text components.


Segmentation Accuracy and Bitrate
To measure the segmentation accuracy of each algorithm,
we used a set of scanned documents along with corresponding
“ground truth” segmentations. First, 38 documents were chosen
from different document types, including flyers, newspapers,
and magazines. The documents were separated into 17 training
images and 21 test images, and then each documentwas scanned
at 300 dots per inch (dpi) resolution on the EPSON STYLUS
PHOTO RX700 scanner. After manually segmenting each of
the scanned documents into text and nontext to create ground
truth segmentations, we used the training images to train the algorithms,
as described in the previous sections.


Computation Time
Table IV shows the computation time in seconds for multiscale-
COS/CCC with three layers, multiscale-COS/CCC with
two layers, COS/CCC, COS, and multiscale-COS/CCC/Zheng.
We evaluated the computation time using an Intel Xeon CPU
(3.20 GHz), and the numbers are averaged on 21 test images.
The block size on the finest resolution layer is set to 32. Notice
that the computation time of multiscale segmentation grows
almost linearly as the number of layers increases. The computation
time of our multiscale-COS/CCC and multiscale-COS/
CCC/Zheng are almost same. We also found that the computation
time for Otsu and Tsai thresholding methods are 0.02 seconds
for all of the test images.

Qualitative Results

illustrates segmentations generated by Otsu/CCC,
multiscale-COS/CCC/Zheng, DjVu, LuraDocument, COS,
COS/CCC, and multiscale-COS/CCC for a 300 dpi test image.
The ground truth segmentation is also shown. This test image
contains many complex features such as different color text,
light-color text on a dark background, and various sizes of text.
As it is shown, COS accurately detects most text components
but the number of false detections is quite large.


CONCLUSION
We presented a novel segmentation algorithm for the
compression of raster documents. While the COS algorithm
generates consistent initial segmentations, the CCC algorithm
substantially reduces false detections through the use of a
component-wise MRF context model. The MRF model uses
a pair-wise Gibbs distribution which more heavily weights
nearby components with similar features. We showed that the
multiscale-COS/CCC algorithm achieves greater text detection
accuracy with a lower false detection rate, as compared to
state-of-the-art commercial MRC products. Such text-only segmentations
are also potentially useful for document processing
applications such as OCR.

plz can u provide total info about text segmentation for mrc document compression and ppt and document

to get information about the topic "TEXT SEGMENTATION FOR MRC DOCUMENT COMPRESSION" full report ppt and related topic refer the link bellow

https://seminarproject.net/Thread-text-s...pid=147070