Open Collections

A transmission system for DPCM encoded monochrome images over a noisy channel

University of British Columbia

Due to the high correlation of neighbouring pixel values of a digital image, DPCM picture transmission has been a well investigated area of image compression. DPCM has found its way into standards such as JPEG and MPEG which have revolutionized the multimedia industry. However, because of the high noise sensitivity of DPCM compressed images, techniques must be developed to reduce the disturbing effects of channel errors. Hence, a DPCM image transmission system over a binary symmetric channel is studied, along with techniques used to improve picture quality. Two concepts which are the foci of the investigation are joint source/channel decoding (JSCD) and post-processing. The DPCM encoded image usually exhibits some residual redundancy. Thus the encoded image can be modeled as a first-order Markov model, and a bit error rate minimizing decoder takes advantage of the source statistics to correct errors. This is a type of constrained joint source/ channel decoding, and does not use added redundancy in the form of FEC to improve picture quality. Usually, side information consisting of source statistics need to be transmitted to the decoder. It is found that an iterative JSCD system which needs no side information has the same error performance as a system which requires error free transmission of the source statistics, but has a larger decoding delay. Furthermore, the JSCD system was generalized to one using a second-order Markov model and the improvement was found to be negligible. JSCD decoding at a very high probability of error, does not correct all of the channel errors; and there remains some very noticeable errors. Post-processing techniques have been applied to corrupted DPCM images and thus it is proposed that the similar techniques be applied to JSCD DPCM images. A thresholding technique is developed to locate the errors and then correct the erroneous symbols. Symbol correction methods which make use of source statistics are compared to a correction technique based solely on the analysis of the decoded DPCM image. Subjective and objective results are presented with the use of the output images and signal-to-noise ratio graphs. These measures demonstrate that the proposed system of a concatenated source/channel decoder followed by a post-processor improves a noise corrupted DPCM image quite effectively.

Thesis/Dissertation

application/pdf

2009-05-21

For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.

http://hdl.handle.net/2429/8026

Electrical and Computer Engineering

University of British Columbia

UBCV

DSpace