Exhaustive Security System Based on Face Recognition Incorporated with Number Plate Identification using Optical Character Recognition

In recent times due to rise in terrorism, people need to live in a safer place where unidentified persons will not be allowed to enter in the premises. Securing of major areas is a vital issue that needs to be addressed for the intelligence and security agencies. At the surrounding of premises, CCTV (Close-Circuit Television) cameras are usually installed to identify the number plate from database by using OCR (Optical Character Recognition) algorithm. This method of security by identifying only vehicle without verifying the person inside it is usually causing serious security issues. Identification of a person is usually done through image processing by using Viola Jones algorithm and acquire the information of the facial components to create a dataset for machine learning. It is imperative to introduce such a system that will be capable to identify the person along with the number plate of vehicle from the stored database. In this research, a comprehensive security system based on face recognition integrated with the vehicle number plate is proposed. The combined information of both dedicated cameras is then transferred to the based station for identification. This system is capable, of securing premises from crime in a more enhanced way.


INTRODUCTION
W ith the increase in terrorism security agencies and high officials are using security systems for securing the territory of the restricted area. In addition to high officials, normal peoples need secure place to live a happy life. Usually special agency guards are hired for the securing of entering and exit gate of the boundary houses. Most of the people use CCTV camera for contineous live monitoring of the boundary and the entering doors to detect the intruders [1]. This footage is sometimes stored in data base for the identification of crime that has held previously. Some people use number plate detection using image processing and the barrier will allow only those cars to enter that are already registered in data base.
as they try to alter facial identity features of the face. Due to pose, expression, makeup variation in 2D (Two-Dimensional) face recognition suffers from poor identification in spite of broad research. The recognition of face using 3D facial surface and shape has increased the discriminating features due to increased dimensionality [11].
Automated security system is of great interest for researchers. Kevin et. al. [12] has proposed an automated security system that operated the entering doors. He designed and implemented an automated system that operated security gate using vehicle license plate recognition by Image processing that extracted character from license plate. A NN (Neural Network) is also used to perform better OCR [13]. Poorvi et. al. [13] presents a robust security system using license plate recognition.
The system consists of four systematic steps that are preprocessing of captured image, extract the region of number plate, its segmentation for identification of character and last character recognition. Memon et. al. 14] proposed an algorithm to identify the characters of Sindhi language using OCR.
It is imperative to introduce such a system that will be capable to identify the person along with the number plate of vehicle. For ideal automated security system person recognition is very important along with the vehicle identification. In recent times, face recognition is very commonly used biometric technique with several applications. Like, in attendance marking, public record authentication, safety and other security systems. Finger print, palm-print and signature is also used for biometric identification but they require user's complete attention and additional time as compared to face identification which requires less user attention (Patil et. al. [11]). So a system that can incorporate number plate identification along with identification of driver will be ideal for the security systems.
This research proposes an exhaustive security system based on face recognition incorporated with number plate identification using optical character recognition. The system has two different dedicated cameras for the better security of the restricted area. The first camera will capture an image of the number plate and the other will capture an image of the driver for further processing. Voila Jones algorithm is used to identify the face and then extract the facial components for centroid calculation thus creating a data set for machine learning and the number plate is recognized through OCR. The combined information is transferred to the base station where it will be compared with the database and identify whether the person driving the vehicle really owns it. If an unknown person tries to enter the restricted area by using an authentic car, the system will instantly block that person at the entrance gate and an alert is generated to inform the concerned authority so they may take any preemptive measures to neutralize the threat if need arises. Fig. 1 represent the software flow control of the proposed system. After system initialization, both cameras located at the main gate will capture their respective images for both face and number plate. As soon as the car triggers the Infra-Red sensor both cameras will capture the images. Camera-1 will capture an image of the person driving the car and apply VJ Algorithm to detect the drivers face in an image, and Camera-2 will capture an image of the number plate. Facial organs are detected from the image acquired by Camera-1 using computer vision toolbox, then the distances between the organs are calculated like the distance between the left and right eye, distance between left eye and nose, distance between left eye and mouth, distance between right eye and nose, distance between right eye and mouth and the distance between nose and mouth. Camera-2 captures an image of the number plate and identifies its number using OCR algorithm.

RESULTS
In Table 1         color in this case. Whole yellow region containing the number plate serial is cropped as shown in Fig. 6(b) and filtered by implementing morphological operation. This image acquired after filtration is converted into binary format as displayed in Fig. 7(a). The detected output is displayed in Fig. 7(b) on notepad.