Download presentation
Presentation is loading. Please wait.
Published byAdela Bruce Modified over 9 years ago
2
Special thanks to our DIP Workshop mentors namely, Sunil Jaiswal, Manohar Kuse, Dushyant Goyal, Gaurav Mittal, Bhavya Agarwal. Thanks also to Ayush Kumar & Nimisha Agarwal for their help in the learning stage.
3
The software segregates the license plate out of the photo of a car taken from a camera.
4
The motive was to find a white expanse of pixels which would be the license plate.
5
We initially turned all other(except white) pixels to black The initial points were noted and length was calculated,and if it turned out to be greater than a minimum amount,we assumed that this was the license plate.
6
This method hugely failed for white cars. It also failed for cars having reflections of lights. It couldn’t differentiate between the headlight and the plate of the car.
7
Due to difference in colors or layers between the plate and the chassis Failure of previous method i.e Thresholding of the image to find the plate.
8
Use of masks to identify all the edges of the plate. The initial points were noted and length was calculated,and if it turned out to be greater than a minimum amount,we assumed that this was the license plate. After the detection of plate,the output image will be produced.
9
The predefined masks gave bad results. Wrong detections of edges. Broken edges of detected plate. No result for slightly tilted plates.
10
Synthesis of appropriate masks for good detection.We used four 5X5 masks all the sides of the plate. Development of the method of “dependent padding” to overcome the problem of tilted plates and to enhance the detected edges.
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.