碩士 / 國立臺中科技大學 / 資訊工程系碩士班 / 103 / This thesis proposes a user integrative image inquiry platform where the users are able to use a smart phone or personal computer and apply application software or a Website applied system to identify different types of fish. The cloud computing service is used to achieve immediate response and provide related information on fish.
Using application software on the smart phone, users are able to choose a Region of Interest (ROI) for the photographed fish images and then upload a sub image of the ROI (or called inquiry image) to the cloud server or upload the ROI sub-image (or called inquiry image) to the cloud server from the Website via the Website application system. Once the server receives the inquiry image from the user it will perform fish recognition. The inquiry image will first process feature extraction such as color or texture and then carry out similarity matching for color or texture between the inquiry image data set and feature extraction. The system will finally return the five top of objects that are the most similar in matching result to the user. The real-time system process will accelerate the similarity matching for the object features. K-means is performed to execute preliminary sifting and carry out detailed similarity matching to accomplish fast and precise image recognition and image retrieval. The system conveys these similar data about fish to the user by illustration, images or video.
To verify the effectiveness and the feasibility of this system we experimented on fish immediate recognition system using a smart phone and personal computer. The experimental results show that the recognition system adopts HSV, gradient and SURF with a recognition rate up to 89%. In this system the cloud server is responsible for the complicated calculations and the large amount of data. The proposed system provides fast and stable efficiency and also reduces the phone requirement, promoting phone endurance to encourage more people use this system.
| Identifer | oai:union.ndltd.org:TW/103NTTI5392006 |
| Date | January 2015 |
| Creators | Bo-Jiun Liu, 劉柏駿 |
| Contributors | Chuen-Horng Lin, 林春宏 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 57 |
Page generated in 0.002 seconds