碩士 / 輔仁大學 / 資訊工程學系碩士班 / 106 / There are numerous kinds of ornamental fish; I have always wondered the species of fish although I have the habit of fishkeeping for years. In general, the features, sizes, and the colors are the key factors to distinguish fish species; therefore, I have developed a user-friendly system to distinguish fish species. This research uses the image processing technique to progress the images of fish, applies the Histogram of Oriented Gradient to extract the features, and further uses the Support Vector Machine to classify the systems. Indeed, the prototype faced some challenges, for instance, the background of the photo was too complicated, and the pattern of fish was not clear enough for the system to process and distinguish. Hence, we use the online Background Burner for complicated photo background, which saves time and enhances the image recognition. On the other hand, we use the exponential transformation method to improve the image contrast, which significantly increases the patterns of the ornamental fish. As a result, the enhancements successfully raise the distinguish rate to about 90%, which is much more accurate than other ornamental fish distinguish systems.
| Identifer | oai:union.ndltd.org:TW/106FJU00396035 |
| Date | January 2018 |
| Creators | KUO CHEN-WEI, 郭 振韋 |
| Contributors | KUO,WEN-YAN, 郭文彥 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 32 |
Page generated in 0.0023 seconds