The Poisonous Mushroom Identify Application Based On Support Vector Machine

Food safety has been a major issue related to national development, social stability andhealth of people, and how to properly solve the problems existing in the field of food safetyhas become an important task facing our government. In the Notice of two of the country andthe published "Document" in2014, the food safety has rose to become the basic state policy.The People Republic of China Food Safety Law, Chapter I Article IX, the State shallencourage and support the implementation of food safety related basic research and appliedresearch, and encourage and support food production operators to improve food safetystandards using advanced technology and management practices.Through literature review, there are many ways to identify the poisonous mushroom.However, due to experimental conditions, materials and equipment or require a lot of relevantexpertise restrictions, there is not an efficient and reliable classification. For this feature, thepaper uses support vector machines in combination with the relevant principles of fungaltaxonomy identification based on support vector machine Toadstool identification methodsand establish a classification model. Analysis and design of mushroom identification systemto achieve an efficient and reliable recognition applications.This work is as follows:(1) Establish mushroom feature coding table. Collating information of mushroom tocompare three identification methods: the Shape recognition method, animal testing andchemical detection method. Choosing fungal taxonomy identification method of Shaperecognition method to identify mushrooms. This method is based on structural characteristicsto judge and comprehensive analysis mushroom. Judgments based on this method of finishingfeatures selected for classification. According to the information coding principles to designcoding system, Code the Structural characteristics and establish the table of classificationcode. Make structural features of narrative can be converted into a computer can processsamples in vector form.(2) Establish recognition model of the Poisonous mushroom. Based on knowledge of theSupport vector machine to build recognition model. Due to the classification of mushroomidentification is a nonlinear problem, through the support vector kernel function to compareand choose the radial basis function as the kernel function. The main question on supportvector machine is solving the minimum convex quadratic programming to use of gradientdescent algorithm.(3) Analysis and design of mushroom identification system. The system uses three-tierarchitecture based on object oriented programming of systems analysis and design. The system provides information on the management of mushroom, and the information can beupdated. According to the established model of mushroom identification, to achieverecognition and manage the mushroom.(4) Experiments performed by the incremental training process to keep the existingtraining results in training process, new incremental samples for training. Because informationis increasing, so that the training process avoids the addition of new samples retrain allsamples. Experimental results, the accuracy of classification of the samples with the additionof increments of continuous improvement, has good classification results.This paper design the application of mushroom identification is based on the SVM.Establish the classification model to identify mushroom and propose optimization strategy formodel training process. Experimental results show that the accuracy and efficiency of themethod. Through combine the relevant principles of SVM and the fungal taxonomyidentification to design and implement the application system. Compared with traditionalToadstool recognition, the application of The Poisonous mushroom recognition based onsupport vector machine has the characteristics of simple and fast speed.