Pattern Recognition Based On Shape Feature Of Insects

In this paper, digital image processing techniques and pattern recognition techniques for image pattern recognition study of insects to achieve the automatic discrimination insects. Main research contents and results are as follows:(1)Insect image pre-processing. This article describes the smooth de-noising, location, uniform, binary adaptive threshold selection, image segmentation, etc. were studied. The results showed that the use of median filtering for smoothing denoising results significantly. By calculating the inertia axis and the image rotation completed a position normalized.(2)Insect image feature extraction. This insect shape features are extracted. Shape features include the same moment, insects, perimeter, area, like roundness, complexity, length width ratio, equivalent circle radius. They do not with the image shift, rotation, scaling and change.(3) The insects form feature of storage. The shape characteristics of the insect to see as a normal distribution, take a number of specimens characterized by expectation and variance of the characteristics of insects, such as the amount of storage to a SQL Server database.(4) insect classification decision module. This paper introduces the fuzzy set theory and fuzzy clustering method used to achieve the feature selection:under the wings of the bright spots Lepidoptera feature extraction relatively stable, and easy to identify the characteristics of the design of the primary classifier, and then use the largest membership-based the principles of fuzzy pattern recognition method of designing a second-class classifier, but without compromising the stability of classifiers under the premise of effectively reducing the classifier recognition time and improve the system efficiency.(5) The system software part of the implementation. This article briefly describes the final part of the implementation of the system software ideas and methods in learning the Visual C + +, the combination of VC can quickly achieve the image data processing and the advantages of being easier to design the user interface, programming of the system software.