Type of Vehicle Recognition Using Template Matching Method

Type of Vehicle Recognition Using Template Matching Method Thiang, Andre Teguh Guntoro, Resmana Lim Electrical Engineering Department Petra Christian University Surabaya - Indonesia

Agenda of Presentation Ø Introduction Ø Object Detection Ø Object Segmentation Ø Matching Ø Experiment Result Ø Discussion and Conclusion

Introduction Overview of this research Ø Purpose: To recognize type of car that enter to park area Ø Three types of car: sedan, van and pickup Ø Using a camera as a sensor to recognize type of car visually Ø Using a simple matching method

Introduction Block Diagram of System

Object Detection Image Preprocessing There are three processes Ø Converting image from RGB to gray scale Ø Histogram Equalization This process will adjust brightness and contrast of image

Object Detection Image Preprocessing Ø Noise removal process using low pass filter with 3 x 3 neighborhood Where is low pass filter kernel

Object Detection Ø It is done in predefined area of the image Ø To detect the existence of the car, the system will substract background image from the image Object detection area and its result

Object Segmentation There are two stages in this process Ø Discard background of the image, so it will show the object only and then morphology operation It is done by substract background image from the image Ø Seek the optimal position of object in the image It is done by calculate the cumulative histogram value for each possible existence object in the image

Object Segmentation Object, object segmentation area and its result

Matching Template matching Ø Compare the image with several template image Ø The image and template image must has the same dimension Ø The image will have similarity value for each template image Ø Identifying type of the car using the highest similarity value Similarity Equation

Experiment Result Ø There are 12 images for template image Ø Similarity value varies from 0. 78124 to 0. 93988 Ø Template matching results high similarity value for image that similar to its template, but it will also happen to other images

Experiment Result

Discussion and Conclusion Ø Template matching method gives a good result in recognizing the type of the car Ø It results high similarity value for image that similar to its template, but it will also happen to other images those are not similar to the template at all Ø For the next step, this method can be combined to feature extraction method like gabor filter to get a better result