The Elevator Group Control System Based On Image Recognition Research

With the development of society, especially in the speeding up process ofurbanization in China, more and more high, super high buildings spring up. Elevator, asa State of the art vertical transportation tool, also has been rapidly developed and wideused. Especially the emergence of elevator group control system brought revolutionarychanges to the elevator industry. However, most of the elevator technology wasmonopolized by foreign large companies. Elevator technology in China is relativelybackward. Elevator technology is an important research subject.Currently the research on elevator group control system is relatively matured, butwas restricted by the imperfect information in elevator systems. This thesis tries tointroduce an image recognition technology into elevator group control system, toprovide more information to the system, and optimize the elevator scheduling.This article firstly use image processing technologies like motion detection onsurveillance video, segment the moving prospects and estimate the number ofpassengers. Through the experimental analysis of existing technologies, we comparedtheir performance, selected mean background model to detect moving target, andmathematical morphology to preprocess image. Combined with the specific scenarios,we improved algorithm for background modeling and estimating the passenger numberin term of real-time, accuracy and others aspects, and achieved good results.After estimated the number of passengers, the elevator scheduling algorithm can beoptimized. This article compare the state-of-art elevator group control algorithms,selected a scheduling algorithm based on multi-objective programming, establish theappropriate target model. Then this thesis combined multi-objective model with theestimated number of passenger to construct a new multi-objective mathematical model.Experiment shows the optimized scheduling algorithm greatly improved the quality ofservice.