Nate Woods

Nate Woods

Computer Vision RSU


This summer, I researched license plate recognition, or more specifically, license plate localization. The process of license plate recognition consists of two main parts. With an input of a live video feed or static pictures, first find a license plate in the picture (license plate localization) then convert the sub-image to plain text. My work is based off of Shen-Zheng Wang and His-Jian Lee’s work in A cascade Framework for a Real-Time Statistical Plate Recognition System. The main focus of their work was to grab as many possible license plates in a picture and eliminate possibilities later using properties of a license plate and conditions of surrounding areas of license plate.


I proposed an algorithm that uses a searching technique to find concurent columns in pictures in order to find a license plate. This searching technique looks for similar sized parallel and vertical lines in a similar region on an image that has been processed to exaggerate vertical changes. More in depth results can be found in my final presentation or final project.

Example 1 Example 2 Example 3