Research On The Application Of Optical Image Processing Methods In Battlefield Exercise Damage Assessment

Destruction assessment based on optical image processing occupies an important position in modern warfare,and how to accurately assess the destruction of targets in the course of actual combat or exercises is of great concern to various military powers.Destruction effect assessment is an important part of military operations,and the key to the assessment is the innovation of optical image processing technology and methods throughout the assessment work.This paper focuses on the scientific problems of optical image processing such as denoising,alignment,change detection and damage assessment,which are systematically studied in the following aspects:First,for the harsh battlefield environment and the influence of the imaging equipment itself,the images are easily disturbed by noise in the generation process.Conventional denoising algorithms can cause damage to the edges and textures of images.According to the denoising requirements of images in the destruction scenario,an improved P-M denoising model is proposed.Firstly,the diffusion function of the P-M model is modified.Secondly,the noise points and dense noise with larger gray values are denoised using an improved algorithm coupled with the P-M model after median filtering.Finally,it is verified by simulation that the improved algorithm can effectively protect the edge and texture information of the destroyed image while denoising.Second,for the problem that the angle and position before and after the destruction images taken by the high-altitude reconnaissance aircraft are difficult to keep consistent,the images before and after the destruction need to be aligned before the image change detection.In this section,an improved algorithm is proposed based on the SURF image alignment algorithm,which uses the BRIEF descriptor to describe the feature points and improves the random consistent sampling algorithm by the geometric constraint algorithm to realize the purification of matched point pairs,and finally the single response matrix is derived and applied to the image to be aligned.The improved algorithm in the simulation experiments decreases the alignment offset by 14.3% on average and the alignment time by 25% on average compared with the SURF image alignment algorithm,which shows excellent noise immunity,accuracy and real-time performance.Third,to address the difficulty of image change detection before and after destruction,this chapter firstly adopts the fuzzy C-mean clustering algorithm for change detection,but the detected results have the problems of too much scatter and low accuracy.In order to further accurately detect the change region,the multi-scale MRF neighborhood information is added on the basis of clustering to propose a multi-scale MFCM change detection model based on the correlation between pixels.The simulation results show that the improved algorithm has a greater improvement in removing the influence of isolated points and improving the accuracy compared with the fuzzy C-mean clustering algorithm,and the analysis of the data collected during the experiments shows that the multiscale MFCM detection algorithm reduces the error rate by 15.04% on average and increases the correct rate by 9.92% on average compared with the fuzzy C-mean clustering algorithm.Fourth,to address the problem that the current theoretical system of damage assessment is not perfect and there is no unified assessment standard,this chapter proposes a damage assessment system based on the combination of geometric and texture features based on the completion of image pre-processing and change detection,and lists in detail the damage assessment influencing factors and the logical flow of damage assessment.Finally,the feasibility of the research is proved by simulation experiments with real damage data.