The authors
constructed two cascade classifiers based on statistical and
Haar features to decrease the complexity of the system and to
improve the detection rate. However, this method will take
much processing time even with low-quality images. Bai et
al. [23] proposed an algorithm for License Plate Detection for
monitoring the highway ticketing systems. Their algorithm
presented a linear filter to smooth the image and to overcome
the influence of light. In addition, vertical edge detection was
used for suppressing horizontal noise. Then, the edge density
was measured and compared with the true plate region
density. A nonlinear filter was applied to remove the narrow
horizontal lines. Finally, a connected component analysis
algorithm was applied to show and to locate the License Plate
features. However, their algorithm works better with a fixed
background and a stationary camera. The most common and
earliest edge detection algorithms are those based on the
gradient, such as the Sobel operator [34] and the Roberts
operator [35]. Numerous previous methods have used the
Sobel operator to extract the vertical edges in Car License
Plate Detections [16], [17], [23], [32], [36]. In this paper, we
proposed the VEDA to extract vertical edges.