Research On Infrared Image Segmentation Based On Swarm Intelligence Algorithm

At present,infrared detection technology has been widely used in power equipment fault detection and become the key technology to realize real-time detection of power equipment.Infrared thermal image of power equipment shows the status of each part of the equipment through temperature cloud diagram and carries out fault diagnosis of the equipment according to the boundary of temperature.Existing literatures only carry out single threshold segmentation for infrared images so as to locate equipment fault diagnosis,but no literatures carry out multi-threshold processing for infrared images.In this paper,the infrared image is divided into several different parts by multi-threshold segmentation,which can not only locate the fault area of the power equipment,but also predict the future operation status of the whole power equipment in different temperature areas.However,the infrared image has the disadvantages of fuzzy,low contrast,unclear texture and impulse noise,which will lead to the decrease of the accuracy of the analysis results.Traditional image threshold segmentation is inefficient,and accurate threshold segmentation results can ensure accurate fault location and more accurate segmentation results of different temperature regions.By referring to relevant literatures on threshold segmentation at home and abroad,it is found that the combination of emerging intelligent algorithm and traditional threshold segmentation algorithm(Otsu,MCE,KSW,etc.)can achieve good threshold segmentation results.This article uses glowworm swarm optimization(GSO),Bird Swarm Algorithm(BSA)and the improvement of glowworm swarm optimization(IGSO)as well as the improved Bird Swarm Algorithm(IBSA)and traditional(MCE and Otsu)threshold segmentation algorithm combining the infrared image threshold segmentation,and compared with the traditional algorithm(Otsu and MCE)and adaptive particle swarm optimization(APSO)combined with the traditional algorithm(MCE and Otsu)segmentation result is compared,The results showed that GSO-Otsu(GSO-MCE),BSA-Otsu(BSA-MCE)and APSO-Otsu(APSO-MCE)achieved better segmentation than Otsu(MCE),and IGSO-Otsu(IGSO-MCE)and IBSA-Otsu(IBSA-MCE)achieved better segmentation than GSO-Otsu(GSO-MCE)and BSA-Otsu(BSA-MCE).First,I will make an deeper study of glowworm swarm optimization(GSO)and bird swarm algorithm(BSA)to understand their operation mechanism and their advantages and disadvantages.I will consult the existing literature about the improvement strategy of this algorithm and the improvement strategy proposed by the similar algorithm,and further analyze whether this improvement is feasible.Secondly,the influence of impulse noise on infrared image segmentation is analyzed.Third,MATLAB software is used to test the intelligent algorithm.Fourthly,PSNR and MSSIM were used as evaluation indexes to evaluate the segmentation effect of threshold value.This research can improve the real-time detection efficiency of power equipment,accurately find the fault point,so as to carry out effective treatment,reduce the economic loss caused by the accident,and ensure the effective operation of power equipment for a long time.