Multi-source Heterogeneous Data Fusion For Intelligent Refrigerator

With the rapid development of information technology,data volume surges,data types become increasingly complex,and data sources become more and more extensive.Large scale data sets a higher demand on database.Large scale data cross and correlation also affect the development of the industry.Data structures in related fields vary greatly and data sources are extensive.These multi-source heterogeneous data contain huge hidden information,which is increasingly playing an important role.How to store data of multiple data sources,multiple structures and connect them effectively becomes an urgent problem.In the above background,a multi-source heterogeneous data fusion framework multi-shdf is proposed in this paper,and on this basis,a data fusion model fd-dfmodel is proposed.Multi-shdf includes multi-source heterogeneous data acquisition,storage and fusion.The raspberry pie is used to collect data during data collection.HBase is used as the data storage center of multi-shdf.On this basis,a data integration scheme is designed based on HBase,and a second-level index scheme based on HBase is designed according to the data characteristics of multi-shdf.The fd-dfmodel hybrid fusion model is proposed,and the decision result after the fusion of feature layer and decision-making layer is taken as the output,which significantly improves the accuracy of object recognition.Taking the fruit recognition in intelligent refrigerator as the application point,this paper proposes a fruit recognition scheme based on multi-shdf frame.The fd-dfmodel was used to construct the fruit recognition model.The recognition model used BP neural network to integrate the results of the fusion of SSD feature vectors based on three network models with the sensor data for decision fusion,and the final recognition results were obtained.The validity of fruit recognition model based on fd-dfmodel was verified,and the experimental results showed that the proposed fruit recognition model based on fd-dfmodel had higher accuracy in fruit recognition of intelligent refrigerator.