Application of computer vision in optimal design of assembled buildings

The shape data of prefabricated building components are closely related to their safety and reliability. To solve the problem of shape energy-saving optimisation, a radial basis function neural network model based on particle swarm optimisation (PSO) considering temperature compensation is studied and designed, and building information modelling technology is introduced as an auxiliary technology for effective management of visual information, which finally realises the energy-saving calculation of building shape dimensions. The results show that the maximum expansion deformation measured by the proposed model appears at the 28th minute, the maximum expansion deformation is 0.11 mm, the error between the model and the actual value is only 0.02 mm and the difference between the monitoring time points is only 3 min. The total energy consumption values of this model are 36.92, 42.15 and 33.58 kWh/m² less than those of the PSO model for three types of buildings. In terms of the total contribution rate of energy conservation, the former are 0.76, 0.88 and 2.94% higher than the latter, respectively. Therefore, this research has effectively improved monocular machine vision technology. At the same time, the energy-saving model of shape with temperature compensation for innovative design has also been effectively verified.