Location and Detection of a Text in a Video

Abstract

Recognizing and identifying the objects present in a video is a challenging task in the area of computer vision. The paper presents a method through image processing activities, to recognize a vehicle passing through a highway. The video of the vehicle is recorded and a frame, which contains the backside or front side of the vehicle, is separated for further processing. This single frame is an image, might have the vehicle object any where in it. A prior knowledge of the structure of the vehicle is used to locate the vehicle in the frame. The number plates of the vehicles normally will have significant features like with the specific background (White or Yellow) and the vehicle numbers are written in a rectangular area with these features is adopted to locate the number plate in the frame. Further segmentation is carried at character level using contour traversal method. A fourteen segment projection method is followed to recognize the characters of the segmented characters. The proposed method is able to segment and recognize the vehicles in the most of the ideal situations.