Research On Digital Image Restoration Forensics Based On MATLAB

Image is an important medium of information transmission.In recent years,the video surveillance system of the public security organs and the commercial and the civilian video facilities have been gradually established.In the detection process of all kinds of cases,images have become a very important source of clues.Image acquisition and storage equipment are various,range from traditional film cameras to digital photography,video cameras,surveillance video and tachographs,smartphones,and even unmanned aerial vehicles,etc.As for practical work,the image from the security surveillance video has become increasingly prominent in the identification and forensics.Security surveillance video is an efficient way to find the required information through browsing.It uses high technology to objectively record all the events that occur in the current surveillance scope,including the process and details of the incident,so that investigators can know the facts or related series-parallel cases and obtain the suspect's height,weight,clothing,appearance and other details,confirm the suspect's confession.At the same time,combining the video surveillance of multiple connected lots can get the track of the criminal suspects,lock the position of hiding and direct the investigators to capture them directly.This is particularly important in the detection of murder cases.Images collected by monitoring usually have the following characteristics.First of all,in most cases,the camera is far away from the observation target,the definition is not high,and the acquired pixel information is less.Second,it is easily affected bythe change of external conditions.For example,the night light,wind,rain and other weather factors will cause image blur.Moreover,the fast moving speed of the target may also make it difficult for the camera to capture a clear image.Therefore,with the combined influence of internal and external factors,such as performance of the monitoring equipment,light,weather and observed target motion,low video resolution,out-of-focus blur and high-light blur and other phenomenon emerge one after another.The quality of the information reflected by the image is too insufficient to meet the requirements of evidence.At present,the case handling department urgently needs to find a more effective image processing method to enhance the quality of the image evidence.Among the numerous image processing software,MATLAB is a high-level programming language and object-oriented with simple syntax rules and a powerful image processing toolbox.Besides of calling MATLAB's own functions for image processing,it can be expanded according to the specific needs of the application.Digital image restoration forensics based on MATLAB is researched in this thesis.This thesis is divided into six parts,corresponding to the six chapters.The first chapter introduces the overview of image restoration.Aim at the characteristics of video surveillance and image work identification in practice,this chapter elucidated formation of video images,common image degradation factors and mathematical model of image restoration.The second chapter expounds the digital image investigation and acquisition.Video image investigation methods and operational considerations are expounded combined with cases process.Chapter three introduces image preprocessing technology,which achieve the image graying and histogram equalization.At the same time,according to the characteristics of different images,selective increase median filter to remove noise to improve the target area and provide good conditions for the follow-up operation.Chapter four elaborates blind recovery of motion blur images.Based on the understanding of license plate images characteristics,use MATLAB to calculate the estimated value of motion blurred image's fuzzy kernel(PSF which contained motion angle and motion scale).According to this parameter,image software and processing algorithms will be chosento realize the image sharpening.Chapter five expounds multi-frame image super-resolution restoration.SRSoftware is used to achieve super-resolution restoration based on multi-frame images in MATLAB environment for low resolution image sequences.At the same time,in order to explain the effectiveness of preprocessing and restoration process,this thesis uses the actual collected images for processing,which aims to provide effective technical support for identifying license plate details,obtaining related criminal information and providing direction and basis for investigation and prosecution.The research of image restoration is still full of challenges.Few images in the actual cases contain only a single or ideal blur.Therefore,there is no perfectly solution suitable for all types of blurs.Although it is impossible to confirm the best recovery method,each algorithm is suitable for dealing with different image characteristics.How to deal with it in a targeted manner is not only the goal of academic research but also the goal of practical processing.It will be an important research direction in the future to achieve superior results by exploring the combination processing of image inspection and identification and mathematical methods.