Microscopic Image Feature Extraction And Recognition Technology For Grain Storage Microbial Analysis

China is the largest country of grain production and storage in the world, only the stored grain in national grain depot is above 100 billion kilograms. Because grain has a variety of microbe and many nutrients in grain are good natural nutrient medium for microbe, once suitable conditions coming, the microbe activities will not only affect food storage security and result in bad quality of grain, but also bring potentially toxic pollution seriously to affect human consumption safety. As a result, developing a kind of scientific, precision and simple detection technology for stored-grain microbe is very necessary and imperious.This thesis is based on existing digital image processing and pattern recognition theory and (?)ocuses on the characteristics extraction and recognition of stored-grain microbe. The main work of this thesis lies on:According as microscopic image is noisy and uneven illumination, this thesis discussed a variety of image pre-processing algorithms to improve image quality, and combine algorithms to meet the different image partition requirements, which make the foundation for follow-up feature extraction. Because stored-grain microbe's patterns are different, this thesis proposes the edge detection by histogram equalization and soft mathematical morphology to reduce noise and enhance image-processing speed by parallel computation. This thesis not only makes extensive and deep research in image recognition, but also research and analysis of traditional methods of feature extraction, and then proposes the method based on moment invariant and multiple classifiers for the microbe which patterns are apparent. In addition, the method based on venation feature and fuzzy theory, for the microbe which patterns are not apparent and object overlap seriously, which can satisfied the stored-grain microbe's recognition by gray co-occurrence matrix of microbe and fuzzy classifier theory.This research has offered the theoretical foundation and technical support for the stored-grain microbe automated identification and detection system, and laid the groundwork for further research.