Development Of Pedestrian Positioning Image Processing System For Airport Disinfection Based On CMOS Camera

Many communicable diseases can be transmitted through air or contact.There are common characteristics of congregation and rapid transmission among communicable diseases,which seriously threaten the health of the public.The airport is a typical large public place,where viruses and bacteria are easy to spread.Therefore,it is necessary to prevent the spread of viruses and bacteria in time.At present,the existing disinfection method is disinfecting each passenger one by one by staff with the disadvantages of labor consumption and insecurity.Therefore,this project developed a set of airport disinfection spraying height measurement software system to realize the automated and personalized disinfection.This topic mainly studies on the human detection and face detection.Firstly,calling the camera is used to collect pedestrian image.Image processing is to detect whether there is pedestrian in the image.If there is pedestrian,the position and size of the pedestrian and the head are determined.Finally,in order to trigger the corresponding height of disinfection device,the height below the pedestrian’s head is calculated.This paper studies the related image processing algorithms and develops the software system.The main research results are as follows:(1)The image acquisition system is built.According to the requirements of the project,a CMOS camera with appropriate pixels is selected,and the layout scheme of the camera is designed.Image acquisition system is developed with MATLAB software,which collect image,store image and call image.(2)The image collected by the image acquisition system is preprocessed.For human detection,image graying is used to reduce computation,image filtering is used to remove noise,Gamma correction is used for improving image contrast.At last,the size of image should be normalized.For face detection,image illumination compensation is for eliminating color offset.(3)Detecting pedestrian adopts the method of statistical learning.Extracting HOG features is the first step,and next step is importing the extracted feature vectors into the trained SVM to classify image.If there is pedestrian in the image,the candidate area of pedestrian is obtained.The height of pedestrian is required to be determined in this project,but the detected candidate area is not accurate.Therefore,the image segmentation of pedestrian candidate area is carried out by using GrabCut algorithm,and pedestrian position and size are finally obtained.(4)Face detection is realized based on skin color features.Establishing Gaussian YCbCr skin color model is based on skin color features,and using the method of threshold segmentation is to obtain skin color candidate regions.Then,the binary image is processed by mathematical morphology to remove holes,burr or other useless information.Finally,non-face regions are removed according to the feature of area,aspect ratio,area ratio and area occupancy.Finally,the face area’s position and size of head are obtained.The experimental results show that the error of disinfection spraying height is less than 3 cm,and the relative error is less than 2.5%.(5)Uncalibrated method is used to measure the height of disinfection spraying,and software system with complete functions is developed by using GUI design module of MATLAB software.The software contains four modules:image acquisition,image preprocessing,pedestrian feature extraction and detection,face feature extraction and detection.The software system uses human-computer interaction interface,and the function of software meets the application requirements.