22-08-2013, 04:52 PM
A Pseudo Lossless Image Compression Method
Pseudo Lossless .pdf (Size: 1.31 MB / Downloads: 24)
Abstract
To present a pseudo lossless compression which
modifies the noise component of the bit data to enhance the
compression without affecting image quality? The hypothesis
behind the study is that the bit data contaminated by noise can be
manipulated without affecting image quality. The compression
method comprises of three steps: (1) to estimate the noise level for
each pixel, (2) to identify those bits contaminated by noise and
replace them with zero, (3) to perform a lossless data
compression on the processed image. The compression ratios are
3.10, 5.24, and 6.60 for CT, MRI, and digitized mammograms
respectively, for the new method which shows a 36.8%, 62.7%,
and 125% increase for the three data sets than original data. The
processed images are evaluated by two image enhancing
techniques: window/level and zoom. They are indistinguishable
from original images. The proposed method demonstrates an
improvement more than 40% in compression ratio than original
image without deterioration in image quality. The qualities of
processed images are the same as compared with those images by
loosy JPEG2000 image compression at compression ratio around
10.
INTRODUCTION
Recent technological advances have made digital radiology
a practical alternative to the film-based system [1]. The fast
retrieval and ease of transmission of digital data makes this
alternative particularly attractive. The essential idea behind
digital radiology is the flow of image information, i.e. retrieval,
communication, display, and archiving of the image data. Both
the speed of data transmission and the space storage
requirements depend on the amount of data. The data
generated by imaging devices of all digital radiological
modalities are massive. Medical imaging data is also increasing
with the latest high-resolution imaging equipment. The PACS
(Picture Archiving and Communication System) is used to
manage and store the large number of images. Large amounts
of data may affect the performance of PACS systems. Data
compression techniques substantially reduce the volume of the
image data generated and thus increase the efficiency of the
information flow.1
MATERIALS AND METHODS
Despite the advance of instrumentation, noise exists in
medical image. The sources of noise are different for each
digital modality. Noise in CT and radiography is due to the
Poisson statistics of x ray photons, to beam hardening, and to
Compton scattering. Noise in MRI is generated by the
presence of patient in the magnet and the background electrical
noise of the system. All these noises are random in nature and
carry no signal information. When an image is contaminated
by noise, the correlation between pixels is degraded. As a
result, the efficiency of image compression is decreased.
DISCUSSION
Pseudo lossless
Several researchers had proposed different "visually"
lossless compression method [15, 16]. They claimed the loss
of image data could be tolerated without affecting the visual
interpretation of an image. JPEG (Joint Photographic Experts
Group) baseline algorithm is the most widely used
compression method [17]. It uses block discrete cosine
transform to decorrelate the image. The information lost in
the compression and reconstruction process will cause
blocking artifacts.
The JPEG 2000 Committee was
established to formulate a new standard based on wavelet
compression [17]. The information loss in wavelet method
will cause blurring in the reconstructed image.
Good et al. [18] used forced-choice just noticeable
difference (JND) studies to compare 4K and 2K laser-printed
posteroanterior chest images. It is demonstrated that although
images are viewed as comparable by radiologists, when forced
to choose the better image, they actually select the higher-
resolution images in 83% of the paired observations. They
concluded that small differences in image quality might be
detectable even in image sets which are considered to be
comparable by subjective assessments.