| Grosvenor Momordica Fruit, the famous specialty of Guilin which has a high nutritional and medical value, is an important economic crops in Guangxi. Now, the cultivate, fabrication of Grosvenor Momordica Fruit has become a good sized industry chain. This research is rooted from the project which is sponsored by Guangxi information and communication technology key laboratory's developing funds, mainly focused on the problem of species lost and degradation in current industry of Grosvenor Momordica Fruit. This research is about the recognition of grosvenor momordica fruit species using image processing and recognition technology based on embedded systems. Not only eliminating faults in the process of manual recognition, but also faster, more efficient and easy to promotion.The main concentration and accomplish of this thesis is:1.Confirmed the hardware and software of the embedded system, proposed a complete scheme on image processing and recognition based on embedded systems corresponding to this project.2.Proposed a scheme for image preprocessing, implemented the image preprocessing algorithm based on both Matlab and ANSI C-language.3.Implemented color, geometric, and shape features extraction from Grosvenor Momordica Fruit image based on both Matlab and ANSI C-languge. And finally chose 13 parameters for image recognition including RGB component, HSI component, perimeter, area and circularity.4.Researched on the feasibility scheme ,designed a Back Propagation artificial neural networks for image recognition,accomplished the trainning of the networks,and accomplish the recognition of Grosvenor Momordica Fruit species.5. Finally, port the entire algorithm to embedded systems. And also port MiniGUI as the graphic use interface, which offers a simple, friendly interface. The system has achieved expected recognition accuracy in the test. This proves it's a effective way for species recognition, and this method may port to other image recognition programs in future.
|
PDF Full Text Request