REAL TIME TRAFFIC SIGN ANALYSIS Presented By- Rakesh Ravaso Patil T CO ‘B’ 12276 Guided By- Ms. P. P. Lokhande
Overview Introduction Traffic sign analysis Color segmentation Edge detection Shape based detection Recognition Binary Thresholding Recognition and Matching using IPP Recognition of traffic signs using FPGA hardware How TSR is works? Conclusion
INTRODUCTION Advanced Driver Assistance Systems (ADAS) Lane Departure Warning Night Vision Automatic Parking Blind Spot Detection Traffic Sign Recognition First used In BMW 7 Series Volkswagen Phaeton
WHY WE REQUIRE THIS? Sleepy driver crashes SUV on Mumbai-Pune Expressway, 7 passengers killed. (TOI, March 5) Human error behind most Expressway mishaps. (TOI, March 5) In 2012, the expressway, witnessed 475 accidents in which 105 people died. MSRDC plan: Trauma Care & Copter Service CCTV Cameras Truck Terminals Reducing U-Turns
TRAFFIC SIGN Possible Sign Type Sign Shape (Border) Colors Triangle, Restricting & Red, Blue, Black Rectangle, Octagon, Warning Circle Information Blue, Red Rectangle Highway Green Rectangle Information Table: Standard Traffic Sign
REAL TIME TRAFFIC SIGN ANALYSIS Detection Recognition Problem facing Illumination affects the color analysis. Occlusion affects the shape analysis. Weather conditions such as rain, snow or fog affect the shape extraction. Physically damaged or changed surface metal of traffic signs affects the recognition.
COLOR SEGMENTATION Fig: Traffic sign and Red/Blue segmented image
COLOR SEGMENTATION-ADVANTAGES Eliminates undesired colors, thus the number of edge pixels in the edge detection process decreases. The complexity decreases since only edge pixels are processed. Fault detections decrease in the detection process. Color segmentation gives information about the border color and the inner color of the sign.
EDGE DETECTION Identifying points in a digital image at which the image brightness changes sharply Fig: Edge image with color segmentation
SHAPE BASED DETECTION Types: Triangle, Circle and Rectangle TRIANGULAR SIGN DETECTION Hough Transform using Slope-Intercept Line equ. y=a.x + b where: x,y are coordinates a is the slope of the line b is the constant parameter… Use of Polar Coordinates instead of Cartesian Coordinates.
TRIANGULAR SIGN DETECTION x.cosΘ + y.sinΘ=rWhere: r is distance between line & Origin Θ is angle from origin to the closest point to line
TRIANGULAR SIGN DETECTION Fig: Edge Image of a Triangular Fig: Detected Lines after applying Traffic sign Hough Transform
CIRCULAR SIGN DETECTION Circular Hough Transform using parametric equation of Circle: (x-xc)² + (y-yc)² = r² Because of Perspective distortion Circular traffic sign may appear as elliptical. (x-xc)² + k.(y-yc)² = r²
CIRCULAR SIGN DETECTIONFig: Detected Circle after applying CHT Fig: Detected Ellipse after applying Ellipse Detection
RECOGNITION A binary image is generated using ROI of the image. Morphological operations are applied to the binary image in order to remove the unwanted pixels. Informative Pixel Percentage (IPP).
BINARY THRESHOLDING ROI is the informative part of the image. Traffic sign consists of only two different colors. One is the informative color of ROI and the other is the background color. Fig: Output of Binarization Process
RECOGNITION AND MATCHINGUSING IPP TRIANGULAR SIGN RECOGNITION Fig: Divided Regions of Triangular Sign
CIRCULAR, RECTANGULAR SIGNRECOGNITION Fig: Divided Regions of Circular and Rectangular Sign
CONCLUSION Automatic traffic sign detection and recognition is an important part of an ADAS. Traffic symbols have several distinguishing features that may be used for their recognition and detection. There are several factors that can hinder effective detection and recognition of traffic signs. The performance of the TSR system can be improved with increasing the number of divided regions.