The Edge Computing Extensive Data Processing Framework and Algorithm for the Internet of Things

Abstract

The paper studies a monitoring pedestrian recognition method driven by big data for edge-cloud collaboration. It extends the original centralized computing to edge and cloud collaborative processing. Firstly, the image boundary node N0 is preprocessed, and the extracted image is expressed in multiple levels. Then, the RGB-D multimodal image learning modeling method is applied to the edge nodes of the network using cloud computing. The boundary node uses the existing learning mode to perform action identification and uploads the identified action information to the cloud to form the final action classification. The method of bone surface fitting and dense trajectories is combined to achieve robust, dense human posture feature extraction. The directed principal component histograms of 3D stereo structures are obtained using dense point cloud data. The features of the spatial and temporal neighborhood 3D gradient histogram are extracted from the apparent texture. Through experiments, it is verified that the proposed method can significantly reduce the shortcomings of traditional centralized algorithms in data transmission and cloud storage and improve the pattern recognition accuracy by 2.2\% based on edge cloud collaboration.