The Application Of Bp Neural Network In Water Microbial Identification System

With the development of digital image processing, pattern recognition and artificial intelligence, neural network BP technology become more and more mature in the field of computer.Its strong adaptability, self-learning and summarizing, the strong antinoise ability, and programming easily to be realized make it particularly suitable for solving complicated problems of internal mechanism, especially for classification problem.The identification of microorganisms in water is a way of judging the quality of water. Water microbial recognition can be drawn from the past artificial recognition by microscopy to automatic recognition of modern computer, the past artificial recognition takes too much time and Recognition errors are high, and the computer recognition overcomes these disadvantages, but the recognition process is not easy. We first need to target the pretreatment after interactive object extraction,and then extract the target characteristics. After the extraction of target characteristics are combined into a description of target feature vector, we need to reduce the dimension of feature vectors because feature vectors are too large to distinguish for the classifier recognition,and so we use a genetic simulated annealing algorithm to optimal combination of feature selection in order to reduce the number of the dimension of feature vectors. Finally we design a good classifier for the identification of microorganisms in water based on this advantages of BP neural network and make it as the classifier of our system. We can constructed a network with weights for the nodes to make a input-output classifier through the training samples for the continuous learning and training for three layer Feed-forward neural networks.The paper first introduces the background and the significance of the research, and simply summarizes the classifier technique.Then this paper introduces the techniques used in Image processing, including pretreatment.watershed algorithm segmentation technology, feature extraction and feature optimization. And then it describes the BP neural network algorithm from the aspect of Calculation theory layer and algorithm layer in detail, and uses genetic algorithm optimizing the initial weights of BP neural network and threshold value.Finally it gives the application of the technology of the BP neural network in the system for the identification of microorganisms in water,and realizes the automatic recognition of the microorganisms in water.