TY  - MANSCPT
AU  - Kamil Zakwan Bin Mohd Azmi
TI  - Development of underwater image contrast and color enhancement methods through optimization algorithm
SN  - THE0010528 (Local)
PY  - 2025///
CY  - Kuantan, Pahang
PB  - UMPSA
KW  - Faculty of Manufacturing & Mechatronics Engineering Technology
KW  - Dissertations
KW  - Universities and colleges
KW  - Theses
N1  - Faculty of Manufacturing & Mechatronics Engineering Technology; Thesis (Doctor of Philosophy) -- Universiti Malaysia Pahang - 2025; Includes bibliographical references
N2  - Improving underwater images is one of the important prerequisites for researchers to explore the ocean, use underwater resources and develop underwater technology. Poor image quality ruins the accuracy of image processing applications. The shapes and colors of underwater objects are extremely difficult to be analyzed accurately due to the phenomenon of color absorption and light scattering. These problems limit the extraction of useful information that can be obtained from underwater images. These undesirable problems need to be addressed thoroughly, causing the underwater image enhancement is an important field to be investigated. Image enhancement techniques offer an effective solution, not only for a better visual experience but also enhance the reliability and robustness of image processing systems to analyze and process the images. There are various underwater image enhancement techniques that have been proposed to address these issues. However, most of these methods inadequately reduce underwater color cast and make a significant improvement to image contrast. The stretching process may become ineffective as the image histogram could be only stretched for a small range of intensity levels. Worst case scenario, the original histogram limits already hit the minimum and maximum dynamic range, resulting in no stretching process and no improvement could be done. Apart from the above-mentioned problems, it is observed that some underwater images are too dark, causing difficulty in interpreting the physical appearance of an object especially when the underwater object is located far away from the camera. In this thesis, three new methods have been proposed to improve the quality of underwater images, namely, swarm-intelligence-based underwater image color enhancement (SUCE), natural-based underwater image color enhancement (NUCE), and intensity-randomized-based underwater image contrast enhancement (IRUCE). The major contribution of SUCE is the reduction of color cast. The adapted gain factors are computed to compensate for the attenuation of minimum color channel, thus reducing unwanted blue-green color cast significantly. Meanwhile, NUCE focuses on improvement of image brightness. Through the incorporation of swarm-intelligence algorithm, the mean intensity values of inferior color channels are adjusted to be closed to the mean intensity value of superior color channel. On the other hand, IRUCE focuses on enhancement of image contrast. An alternative approach of histogram modification has been proposed to overcome the limited stretching issue of conventional global histogram stretching called as intensity-randomized technique. This method is developed to ensure that image intensity is properly distributed across all intensity levels. This technique is designed to fuse between under-enhanced and over-enhanced images. By utilizing the random values which are drawn from normal distribution, the image intensity is well-distributed across all intensity levels. The effectiveness of these proposed methods are tested with 300 underwater images and the results are compared to the other state-of-the-art methods. The results show that these proposed methods can improve underwater image quality significantly, while outperforms the other methods. The results show that IRUCE achieved the best performance with entropy = 7.787, AveGrad = 10.316, Sobel Count = 36913, and UIQM = 1.590, confirming that the proposed algorithms significantly improve underwater image quality and outperform existing methods. Beyond academic contributions, the findings offer practical impacts for real-world applications such as marine biology research, underwater exploration, search-and-rescue operations, and industrial inspections, where clearer images are critical for accurate decision-making and operational success
ER  -