This paper presents a novel computer vision technique of intelligent recognition of safety in construction sites. The proposed technique relies on a new methodology for Neuro-Fuzzy Singular Value Decomposition (NFSVD). This methodology utilized unmanned safety worker’s wears. Where within a construction site entranced a multi-camera is used and fixed on a special frame inside the security Cabin. Before the workers allowed for entering the construction site, an automatic detecting, and an intelligent matching system has the ability to recognize the worker’s dress shapes and colors through a multi snap image. To improve the success of this methodology, the proposed technique empirically investigated, analyzed and verified to check the image frame quality, accept and reject indication, real-time complexity, diffusion, confusion and PSNR. The results performed by this technique demonstrated the robustness and reliability of this methodology for safety in the construction field. © 2019 NSP, Natural Sciences Publishing Cor.

View