PEDESTRIAN AND OBJECT DETECTION SYSTEM IN VEHICLES FOR SAFETY DRIVING

Transcription

1 PEDESTRIAN AND OBJECT DETECTION SYSTEM IN VEHICLES FOR SAFETY DRIVING J. Ram Prabu and S. Abinaya Department of EEE, Kumaraguru College of Technology, Coimbatore, India ABSTRACT In this paper a new image processing approach is being proposed for detection of pedestrians and objects in roads while driving. The camera is placed in vehicles and image processing functions are carried out and message is sent to the driver by using a controller fitted in the vehicle. The message to driver is to take left,, right or slowdown based on position of object or crossing pedestrian on road. The message sent to driver is either audio message or text message, audio message will be through speakers in helmet and text message will be displayed through LCD devices. As a whole this is a safety system for two wheeler and four wheeler drivers by avoiding road accidents. I. INTRODUCTION The problem is that 50% of the road accidents is due to Pedestrians walking at the foot path crossing the lane and entering in the road and crossing the roads during peak traffic time, unaware of the vehicles passing by[1]. This causes accidents in roads and at the extreme end due to uncontrollability by drivers may lead to injuries and fatalities. The rate of accidents is increasing every year in all countries due to over speeding, pedestrians crossing the busy road and due to some kind of objects in the road. Cyclists and pedestrians account for more than half of road accidents and fatalities in the country. Governments in various countries like America, Africa, Asia, Europe are taking suitable measures to avoid and limit road accidents. The numbers of pedestrian deaths are more in china and India which is given by the survey by WHO. Fatality Rate due to pedestrians is increasing by 8% every year[3]. According to the report of fatality, children of age 15 or below and old people of age 70 and more are accounting more to the pedestrian crashes and the pedestrian fatality rate of male are double or more than female pedestrians. This shows that men walk more than women. Most of the accidents happens due to brakes or when drivers does not take desire actions. So by early detection of pedestrians and object in the road by image processing algorithm like background subtraction can avoid road accidents and to decrease the fatality rate. II. SYSTEM OVERVIEW A way to capture image from camera and do image processing functions based on captured image using a processor and then transferring message to the driver to take desired action based on message through audio device or LCD interfaced with the controller in the vehicle. This is a safety system for two wheelers and four wheelers and basically an alert system for drivers to avoid accidents System Description The description of the system is explained with the below block diagram Vol. 6, Issue 6, pp

2 CAMERA (IMAGE /VIDEO) VEHICLE PROCESSOROR (IMAGE SUBTRACTION + RECT ROI) RESULT TO MICROCONTROLLER BASED ON OBJECT AND DISTANCE Figure 1. Block Diagram of system functioning. A camera is placed in vehicle and the image is grabbed for processing by the image processing algorithm based on the distance of pedestrian and objects the result is sent to the microcontroller as given in fig1. The DSP processor is used for functioning of image processing algorithms and the microcontroller is used in the next part of message transmission to vehicle riders. The Input/output pins is used for interfacing LCD devices in microcontroller, the LCD display is placed in front of the driver and text message is displayed in LCD device such as to take left, right and to slow down. To send audio message sample rate microcontroller is used and a DAC converter which converts digital signals to analog signals and the antialiasing filter which restricts the bandwidth of the signal and attenuates higher frequencies of the signal. And also prevents out of band noise and various other interference of signal. MICROCONTROLLER I/O PINS SAMPLE RATE MICROCONTR- OLLER DAC & ANTIALIASI NG FILTER AMPLIFIER LCD DISPLAY(TEXT MESSAGE) SPEAKER (AUDIO MESSAGE) Figure 2. Block Diagram of Message Transmission to drivers The amplifier amplifies the signal and its connected to speakers. Speakers or Head phones can be used. For two wheeler drivers, headphones or speakers can be placed at helmet which makes all drivers to wear helmet regularly Motion Detection The captured image from camera in the vehicle is sent to the processor where image processing functions are carried out. The background image is chosen first from the set of image sequences. Then each and every image is subtracted from the background image. First the road surface will be continuously monitored and when a object or pedestrian occurs immediately the distance of the object from the driver is manipulated and result is sent to the controller as in [2].The method does two steps motion detection and object tracking Background Subtraction This method is carried out by subtracting intensity values of each pixel from the current image and background image. The R component, B component and G component in an image is subtracted individually and averaged finally to get the background results. The image subtraction formula from[2], A(x,y,c) = abs( Xn(x,y,c)-X0(x,y,c) ); for all x, y, c B(x,y)=(A(x,y,1)+A(x,y,2)+A(x,y,3))/3; where Xn and X0 are the current image and the background image respectively, x and y are the rectangular coordinates, c is the color index: 1 for red, 2 for green, 3 for blue, B(x,y) is the subtracting result Vol. 6, Issue 6, pp

3 2.4. Masking The Region of interest is taken as Rectangle ROI as in [4]. Masking in image processing filters only a part or the required portion of image as in [2].Basically masking is done to remove noise in an image. The noise can be reduced by convolving original image with the masked image. The difference image is convolved with an average mask which takes a local average computation for each pixel. Then, it is converted into a binary difference image by thresholding. The Gaussian noise can be effective reduced by this simple process and the changing regions appear from [5] Flow chart of the proposed system The proposed system is represented through flowchart and the decision is based on algorithm. Figure 3. Flow Chart for the proposed System 2568 Vol. 6, Issue 6, pp

4 2.6. Simulation Results The results by doing background subtraction of captured images from roads and with rectangular Region Of Interest. Figure 4. Sample Road Image Figure 5. Object in Road Figure 6. Object detected and direction to take by driver without hitting object with Rectangle ROI Figure 7. Sample Road image with pedestrian. Figure 8. Pedestrian Detected on Road Based on the detected position of object, the information will be sent to driver to take the right path Decision making The three indications made are Take Right, Take Left and Slow Down as given in Fig3. The image frame is divided into two halves left and right. If the object lies to the left of the frame, the indication is shown to take right. If the object lies to the right of the frame, the indication is shown to take left. If the object lies at the centre of the frame as given in Table Vol. 6, Issue 6, pp

5 Table 1.Decision Making Table Object/Pedestrian Detection at Right Left Centre LCD Display Audio Message TAKE LEFT TAKE RIGHT SLOW DOWN III. RESULTS AND DISCUSSION The image processing operations carried out with background subtraction and Rectangular region is taken as the region of interest. The algorithm detects obstacles at the range of 20 to 30 metres from two wheeler. Based on the speed of the vehicle and the distance of the object detected, the response time can be calculated with the formula, Time = Distance/speed The Table 2, below shows response times at different speeds, Table 2.Response Time at Different Speeds Speed in kmph Response Time in seconds 20 m 30 m IV. CONCLUSIONS The system was efficient in detecting pedestrians and objects on the road. This method can detect motions and do image subtraction and reduce noise by masking method. This helps drivers to ensure that there are no objects and pedestrians crossing the road. This algorithm detects not only Objects and pedestrians but also animals like dogs and cats which cross the road at unexpected times. The implementation of this system ensures enough safety for drivers of both two wheelers and four wheelers. V. FUTURE WORK In Future, this image processing Algorithm can be extended to detect potholes and speed breakers on roads, which ensures enough safety system for to wheeler and four wheeler drivers. Further it can also be extended to detect potholes and obstacles at night times by eliminating shadows. REFERENCES [1]. Shamo Hisaka Metev and and Shunsuke Kamijo onboard wireless sensor for collision avoidance and pedestrian detection,14 th international IEEE conference on intelligent and Transportation Systems, Vol. 6, Issue 6, pp

6 [2]. J. Breckling, Ed., A background subtraction based video object detection and motion tracking,,horng@kmit.edu.tw,2011 [3]. Seiffert Ulrich, Wech Lothar, Automotive safety hand book SAE international,united states of America,2009 [4]. Cho,M, K,Ku, Y.shi and Kanagawa, Ahuman interface design of collision detection and warning system,2001 [5]. Broggi A,Ceni P,Jung HG Anew approach to urban pedestrian detection for automatic braking,ieee transaction Intell transport syst.,no.10.pp ,2009 [6]. James A.Misener*etal: A path investigations in vehicle road side cooperation and safety:a foundation for safety and vehicle infrastructure integration Research,IEEE ITSC,Toronto,2006. [7]. Analysis on car-bicycle Rear end Collision (Japnese),Arm,ITRADA,2010,pp.2-4. [8]. Analysis on car-to-pedestrian Accidents Using Micro and Macro accidents data(japnese),ann.itrada,2009,pp.16 [9]. S.Kamijo, K.Fujimura, Y.shibayama, pedestrian detection algorithm for onboard cameras of multi view angles, G.proc of 2010 IEEE Intelligent Vehicles Symposium*pp ,2010. [10]. Y.Huang obstacle detection in urban traffic using Stereovision, proc. of the 8 th international IEEE conference on Intelligent Transportation System,Vienna,Austria,sept,2005Intelligent Transportation Systems, Vienna, Austria, Sept., 2005 AUTHORS J. Ramprabu was born in Coimbatore, in India. He received his B.E. (Electrical and Electronics) degree from Sri Ramakrishna Engineering College, Coimbatore, India. He received the M.E (Applied Electronics) degree in Kumaraguru College of Technology, Coimbatore, India. He received the M.B.A (International Business Management) degree from Anna University of Technology, Coimbatore, India, in He joined the Department of Electrical and Electronics Engineering, Kumaraguru College of Technology, Coimbatore, since His current research interests include Applied Electronics, Embedded system. Mr.Ramprabu is a Member of the Indian Society for Technical Education, System society of india and published many papers in international journals and conferences. Abinaya is presently doing her master degree (Embedded Systems) in kumaraguru College Of Technology, Coimbatore, India Vol. 6, Issue 6, pp