Article Preview
TopIntroduction
Vehicles have been used widely in every area of production and life of people, and the license plate number is a simple and an effective way for identifying vehicle, it is unique information for every car. With the total number of vehicles increasing quickly, traffic violations will appear more frequently in the public traffic, such as running the red light, over speed, etc.
The license plate detection is considered as an important research topic of vehicle license plates (VLP) recognition system and an extra critical part of modern intelligent transportation system; it is widely used in many applications such as detection of speeding cars, security control in restricted areas, unattended parking zone, traffic law enforcement and electronic toll collection. Last few years have seen a continued increase in the need for and use of VLP recognition. Research on Vehicle License Plate detection and recognition has recently become an active topic (Chen, Yang, Zhang, & Waibel, 2002; Escalera, Armingol, & Mata, 2003; Escalera, Armingol, Pastor, & Rodriguez, 2004; Fang, Chen, & Fuh, 2003). Thus, many vehicle plate recognition systems are developed by researchers (Rahman, Badawy, & Radmanesh, 2003; Wang & Lee, 2003; Chang, Chen, Chung, & Chen, 2004; Zhang, Jia, He, & Wu, 2006; Naito, Tsukada, Yamada, Kozuka, & Yamamoto, 2000). Nonetheless, robust detection of different types of license plates in varying poses, changing lighting conditions and corrupting image noise without using an external illumination source, still present the challenges of the existing VLP recognition systems. In the next we introduce some related work of this field.
The numerous proposed VLP detection systems are based on different properties. Some techniques use simple rules based on deterministic methods while others employ elegant training based classifiers.
For example, a support vector machine (SVM) is used in Kim et al. (2002) to analyze the color textural properties of VLPs. The color values of the raw pixels that make up the color textural pattern are fed directly to the SVM, which works well even in high-dimensional spaces. Next, LP regions are identified by applying a continuously adaptive mean shift algorithm (CAMShift) to the results of the color texture analysis. In Jia, Zhang, and He (2007) work, the mean shift is used to filter and segment a color vehicle image in order to get candidate regions. These candidate regions are then analyzed and classified in order to decide whether a candidate region contains a license plate or none.
Another approach uses gray level morphology (Hsieh, Yu, & Chen, 2002). This approach focuses on local appearance properties of license plate regions such as brightness, symmetry, and orientation. Candidate regions are compared with a given license plate image based on the similarity of these appearance properties. Alternatively, Wu, Lei, Chan, Tong, and Ng (2005) combine morphological operations and a projection searching algorithm for vehicles in Macao. In particular, the projection searching algorithm is used to detect region of the characters in the license plate through vertical and horizontal projections. We notice that the license plate styles found on vehicles in Macao are very much similar to the ones found in Malaysia and this can be a solid reference.
Besides, Kim et al. (2006) proposes, a method based on edge extraction for license plate localization in images taken in poor lighting conditions. Their method is composed of two steps: The first step involves the search of candidate regions from the input image using gradient information while the second one, determines the plate area among candidates and adjusts the boundary of the area by introducing a plate template.