Being one of the three methods for judicial expertise, image data is vital to the criminal detection and judicial expertise. However, limited by the various environments or equipments, quality of the surveillance videos obtained from the crime scene is usually under average, making amounts of the videos unqualified for the effective evidence in litigation process. Facing the increasing number of "poor quality" image data dilemma, to provide a convenient and effective access to image analysis and management, as well as a flexible and versatile access to the quality improvement in different systems, is of vast importance to help the investigators to discover more clues and evidence to crack down on the crime. Aiming at the surveillance videos and images of public security forensic investigation, this thesis proposes the schemes of mass image data parallel processing and distributed storage, and realizes the networking image analysis service platform oriented to the multi-system as well. This platform ensures that different systems can retrieve relevant services according to the demands, together with improving the mass-damaged-image’s quality by means of real-time or non-real-time recover as well as subjective processing to assist the investigation. Besides, it can also improve the retrieved image’s quality of crime scenes and effectively store the evidences, thus strongly fighting against crimes. This thesis mainly studies and solves three practical problems: 1)Realizing and integrating the surveillance video oriented processing algorithms 2)Realizing parallel processing and distributed storage of massive surveillance video 3)Integrating disperse image processing functions to form operable standard services With regard to the problems above, this topic is intended to investigate the demands of networking image analysis service platform. It researches image analysis and storage based on Hadoop as well as the surveillance image processing arithmetic encapsulation and release management based on Web Services for the sake of meeting the requirements such as great capacity of data, large visitor volume and complex arithmetic.
修改评论