الفهرس 
4 A Quantum Mechanics-Based Framework for Image ProcessingOnly 14 pages are availabe for public view
 |  | from | 16 |  |  |
| | | from | 16 | | |
Abstract
Quantum mechanics provides the physical laws governing microscopic
systems. A novel and generic framework based on quantum
mechanics for image processing is proposed in this thesis. The basic
idea is to map each image element to a quantum system. This
enables the utilization of the quantum mechanics powerful theory in
solving image processing problems. The initial states of the image
elements are evolved to the nal states, controlled by an external
force derived from the image features. The nal states can be designed
to correspond to the class of the element providing solutions
to image segmentation, object recognition, and image classi cation
problems. In this work, the formulation of the framework for a
single object segmentation problem is developed. The proposed
algorithm based on this framework consists of four major steps. The
rst step is designing and estimating the operator that controls the
evolution process from image features. The states associated with
the pixels of the image are initialized in the second step. In the third
step, the system is evolved. Finally, a measurement is performed
to determine the output. The implementation of this algorithm
on a quantum computer is introduced. The presented algorithm is
tested on noiseless and noisy synthetic images as well as natural
images. The average of the obtained results is 98.5 % for sensitivity
and 99.7 % for speci city. A comparison with other segmentation
algorithms is performed showing the superior performance of the
proposed method. This algorithm is then extended to solve an important
biomedical image processing problem which is blood vessel
segmentation in retinal images. First, images are preprocessed for
enhancement. Second, features are generated and extracted from the
image. Finally, the quantum mechanics based algorithm is applied.The algorithm is tested on the publicly available DRIVE database.
The results of this method on the average are 80.29 %, 97.34 %,
and 95.83 % for sensitivity, speci city, and accuracy. These results.are compared with the state-of-the-art methods. The application of
the introduced quantum based framework to image segmentation
demonstrates high eciency in handling di erent types of images.
Moreover, it can be extended to multi-object segmentation and utilized
in other applications in the elds of signal and image processing