碩士 / 國立中山大學 / 海洋環境及工程學系研究所 / 101 / Abstract
Using acoustic telemetry techniques to get underwater environmental information is a normal procedure. Acoustic instrumemts can offer large-scale area coverage and high-resolution imagery in relatively short period of time. It can improve the safety and efficiency in underwater works by non-contacting ways. The scanning sonar can detect and locate underwater targets effectively. In addition, it can monitor fish swimming behavior within the scanning area.
The interpretation and identification of information from the scanning sonar imagery are mainly depended on visual observation and personal experiences. Recent studies tended to improve the identification efficiency by using numerical analysis methods. This can reduce the error that cause by the differences of observer’s experience as well as by extended time of observation. This study was intended to develop a fully automatic sonar imagery processing system for the identification and characterization of both static and moving objects.
The major procedures of the automatic sonar imagery processing system are as follows:
(1) Image Acquisition.
(2) Intensifying the image of target
(3) Feature Extraction：grey level co-occurrence matrix(GLCM)
(4) Classification：unsupervised Bayesian classifier.
(5) Object Identification：by characteristic feature (i.e., Mean value).
(6) Object’s Status Analysis：object’s circumference, area, center of mass and quantity.
This study used the sonar images collected at the Shao-Chuan-Tou yacht wharf site in Kaohsiung City as a case study for static target investigation and at the Lu-Jhu milkfish pond site in Kaohsiung City as a case study for moving target investigation. The image characteristic functions employed include one set of first order parameter (i.e., mean value) and two sets of second order parameter (i.e., entropy and homogeneity). Five sonar images were used to find out the optimum texels and sliding size. The identification efficiency of the system, in terms of the producer’s accuracy, is 84.6%. Based on this ground-truth investigation, the optimum texels size was concluded to be 12×12 pixels with sliding distance of 4 pixels.
Regarding static object investigations, the application of the proposed system included the integration of separate consecutive local images to build a global image mosaic. This procedure can be used to evaluate underwater environment in helping underwater search and rescue activities, in detecting environmental changes over time and in combining with GIS for further investigation. In the field of moving target investigation, swimming behavior of fish could be evaluated. According to the analysis results, the swimming activity of milkfish followed a regular pattern. And milkfish in the scanning area showed high variability of swimming activity and biomass accumulation during the pellets tossing period. Automatic detection of moving object categories using a single characteristic function of mean value is feasible. In this case, objects with mean value larger than 31 represents positive identification of moving target catagory (i.e., milkfish). Non-target categories such as seabed were identified if the mean value is smaller than 31.
| Identifer | oai:union.ndltd.org:TW/101NSYS5282033 |
| Date | January 2013 |
| Creators | Kai-Yin Tian, 田楷寅 |
| Contributors | W.M. Tian, 田文敏 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 228 |
Page generated in 0.0139 seconds