The continuous monitoring of beaches and structures near, or in the surf-zone, has become a vital task from an environmental and economic standpoint. Specifically, the Durban beaches are continually nourished with sand that is dredged south of the Durban Harbour mouth. Constant monitoring is essential to control the erosion and accretion of these beaches. Currently, infrequent and labour-intensive manual surveys are being used to fulfil this task. This dissertation describes the techniques used to process and validate surf-zone video images to extract information, which is equivalent but more frequent and cost effective to that obtained using manual surveying methods. The emphasis is on the extraction of accurate hourly waterlines by analysing video images obtained from Coastal Imaging stations located at Addington Beach and North Beach in Durban, South Africa and the measurement of rip currents in the surf-zone from video images together with the validation of these results. The implementation of a neural network incorporating non-local image pixel data is more reliable than previously used methods, for example, grayscale thresholding or the delta-discriminator. Grayscale thresholding relies on the difference between the water and the sand pixel intensities. The delta-discriminator incorporates hue-saturation-lightness (HSL) values to improve the discrimination between water and sand. These methods could not, however, overcome the problem of varying seasonal and diurnal light intensities and require the prior choice of threshold values. The use of non-local data, such as the average RGB-values, in addition to individual pixel values as inputs to a neural network is shown to give better results in changing ambient lighting conditions. Wavelet and Fourier analysis of the temporal nature of the breaking waves in the surf-zone is also used to calculate more accurate waterlines. These extracted waterlines are validated by data collected from manual surveys. The use of Digital Correlation Image Velocimetry (DCIV) has been proven as a successful method in tracing velocity flow fields in a fluid medium. Typically, the medium is populated with distinct seeds that are tracked through successive video frames using a cross-correlation technique. It is shown that DCIV can be used in surf-zone images to measure surface rip currents by tracking the structure of the advected foam. This technique was validated by simultaneously tracking a large number of yellow corks scattered on the water surface. / Thesis (M.Sc.)-University of Natal, Durban, 2001.
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:ukzn/oai:http://researchspace.ukzn.ac.za:10413/3796 |
| Date | January 2001 |
| Creators | Naicker, Jaysen. |
| Contributors | Alport, Michael J. |
| Source Sets | South African National ETD Portal |
| Language | English |
| Detected Language | English |
| Type | Thesis |
Page generated in 0.0014 seconds