Download files
Abstract
Digital images captured under adverse environments can be vulnerably degraded in their capacities to convey adequate amount of information to the viewer or computer-based processes. This research is focused on two primary types of degradations - images with loss of contrast and colour vividness. Efficient algorithms are developed in overcoming shortcomings inherited with available state-of-the-art approaches.
Manipulating the intensity distribution is one of the popular methods that have been widely employed in image contrast enhancement. However, this conventional procedure usually generates undesirable artefacts and causes reductions in the information content. An approach based on expanding and compressing the intensity dynamic range is proposed in this thesis. As a main category of degraded colour vividness, hazy images are utilised to examine the adaptability of developed algorithm. The experiment verifies that no satisfactory results can be achieved merely through contrast enhancement algorithm. Therefore, an effective saturation enhancement operation followed by histogram specification is proposed for image de-hazing.
However, given an input image severely hazed, the developed algorithm is incapable of achieving attractive results. Therefore, research specifically focused on image de-hazing is conducted. Among all the available methods, the Dark Channel Prior based algorithm has been regarded as the state-of-the-art in recent years. Despite of the satisfactory performance, it is inherited with shortcomings of introducing colour distortion and demanding for further transmission refinement. Therefore, an algorithm to realise image de-hazing from the perspective of noise filtering is proposed in this thesis.
Additionally, an approach named as Image De-hazing Based on Polynomial Estimation and Steepest Descent Concept is proposed to derive pixel-wise transmission that does not require any further refinement. Moreover, image de-hazing procedures based on the steepest descent concept are adopted so that the objective of saturation enhancement under the minimum hue change constraint is achieved. Experiments are conducted on a variety of input images, the results of which are analysed both qualitatively and quantitatively compared to the available state-of-the-art methods. Computation efficiency is another critical factor that has been taken into consideration in the evaluation of the algorithm performances.