**Project Title: Criminal investigation management system based on criminal recognition using CCTV footage**
**Main objective**
This research study is designed to detect and identify criminals in less time with high accuracy, better than the existing methods, using their faces, figures and behaviors followed by the image enhancement process.
**Main Research questions**
Criminal identification process needs a lot of resources such as investigators or police officers, reviewing CCTV footage, identifying the suspect’s face, stature , and manual checking of multiple documents which are time consuming and expensive. These manual identification processes may cause errors in some cases, since it can mis-identify innocent persons who have identical faces/features of the criminal appearing in the footage. To minimize the disadvantages explained above , a computerized identification system can be utilized. Currently there is no such a system in police and at CID of Sri Lanka.
**Individual research question**
**Component 1 – IT18109290 | Fernando W.S.D**
• As a society where criminal cases are recorded every day, it is important to have some better criminal investigation tools. These tools should be fast, reliable, and accurate. Everywhere in the world, CCTV cameras and CCTV footage are used often in the criminal investigation process. To get an accurate result, that CCTV footage must be of good quality.
• In Sri Lanka, the amount of surveillance camera usage is increasing rapidly. But the question is whether people are satisfied with those devices. This is where image and video enhancement would be useful because if the footage is not in a useful state, we can use some form of enhancement method to make it useful.
• There are many image enhancement methods available. We use most of the image enhancement methods, analyze those methods with the image then choose the best enhancement method to enhance the image.
**Component 2 – IT18052152 | Perera H.G.G.M**
• The number of unusual activities identified by CCTV cameras is growing by the day, and these incidents are being reported to authorities on a large scale across the world. As a result of the use of paper-based systems, police officers must devote a significant amount of time and manpower to detecting abnormal activities in video footage.
• Because the collected data can be comprehensive and authorities don't know the precise time when the abnormal activities occur, traditional techniques require authorities to watch the CCTV footage data over and over for hours at a time. As a result, the concentration of authorities on each case steadily decreases over time.
• Because of the lack of human vision precision in detecting small objects as small as a handgun among a group of people, detecting dangerous situations in extremely dense and complex crowds in CCTV footage is impossible and time consuming.
**Component 3 – IT18028188 | Fernando K.P.P.E**
• Using manual methods to identify the criminal and to continue the investigation is time consuming and it costs higher cost, and it may cause errors in some cases , since it can mis-identify persons who have identical faces/features of the criminal appearing in the footage.
• Recognition of the criminal is challenging due to the variations of head pose and viewing angles of the person . So, identifying the suspect with a higher accuracy level is a cumbersome process.
• Identify the suspect using manual methods are time consuming and it cost higher amount to continue the investigation process.
• Difficult to identify human body poses with high variability in a movement such as walking, running, etc. For example, if a person walks along a circular direction, then the location of the head, legs and arms varies with high variability.
• Difficulty in recognition, when a suspect wearing clothes like jackets, coats can affect the measurements of the figure in identification process.
**Individual Objectives,**
• Component 1 - Image Enhancement
• Component 2 - Suspect identification with Abnormal Behavior
• Component 3 - Suspect identification with Face Recognition
• Component 4 - Suspect identification with Figure Recognition
