Alexandria Engineering Journal (Oct 2021)
Application of network protocol improvement and image content search in mathematical calculus 3D modeling video analysis
Abstract
This article connects mobile terminals through improved network protocols, and connects monitoring devices through improved network protocols in the local area network. At the same time, it can convert mobile terminal requests into requests that comply with improved network protocols and forward them to monitoring devices. Method of three-dimensional feature extraction based on dynamic data. First, normalize the coordinates and scales of the three-dimensional calculus model, and use two kinds of data (voxel representation and pixel representation) to deal with video object recognition. Then construct a convolutional neural network as an input to the network, extract visual information and geometric information, and use deep learning to improve the recognition ability of single-modal features. Based on the convolutional neural network, the geometric descriptor and the view descriptor of the three-dimensional calculus model are extracted separately, and the two feature descriptors are fused in a network to find the association between the patterns. After forming the fused feature descriptor, it is applied to micro Classification and retrieval of integral 3D features. The multi-feature fusion layer can not only learn more distinguishing features of the two descriptors, but also supplement relevant information between the two descriptors. Compared with using these two representations alone, this method produces a much better classifier and improves retrieval efficiency.
