Research On Water Turbidity Detection Based On Image Recognition

With the rapid development of computer vision theory and application research,the superiority of computer vision technology in image processing applications is becoming more and more prominent.The traditional method of image processing is not only time-time-saving and not highly accurate,but the regeneration neural network can solve the problem,so the reel neural network has become the focus of current research.Water quality testing plays an important role in pollution control.It is an important basis for correctly judging and evaluating the water environment and the comprehensive utilization of wastewater,the turbidity of water is also an important index to determine whether tap water can be consumed,and turbidity measurement also plays an important role in industrial,medical and scientific research.First of all,this paper summarizes the relevant principles and techniques of reel neural network(CNN)and its application in image recognition,and explains its overall structure through a simple retication neural network model,which reflects the training process of the network.Secondly,the neural network model of turbidity detection is designed.By comparing with the commonly used VGG model architecture,the main structure finally adopts the ResNet model architecture of the resnet neural network,combined with the specific requirements,after experimental testing,the res Net model architecture with the internal structure of the resnet is 34 layers,and the model architecture is designed.Third,the water turbidity detection,the completion of image acquisition,image processing and the system corresponding software and hardware specific design,the design module workflow is introduced,and carried out an experimental verification process,after verification can be learned that the use of the scheme reached a high accuracy rate.Finally,considering the convenience and comprehensiveness of the application,the turbidity detection system of constitutive neural network based on embedded platform is designed.Firstly,it introduces the construction of embedded platform hardware environment,embedded selection and the composition of peripheral hardware equipment.Then the software design of the overall overview,and finally completed the program design of each module.Compared with the traditional turbidity recognition method,the scheme not only has a lower cost,strong real-time,but also has a high accuracy,so the new detection method has good application value.Figure [59] table [3] reference [82]...