.png){kind=icon}
.png){kind=icon}
Return
License plate localization using kernel search multiwavelet decomposition
.png){kind=icon}
PRE
AI
DOI：10.1007/s11042-023-14570-3.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
According to global survey reports, license plate recognition (LPR) provides rich information in approximating the traffic conditions of urban arterials is an emerging data source. Several researchers studied and investigated about segmenting, extracting and classifying the license plate and their approaches do not provide accurate extraction in the different weather condition (night, day, rainy, cloudy etc.). In this research work, a novel feature extraction technique called as Kernel Search Multiwavelet Decomposition (KsMWD) is proposed for license plate detection. By computing the dissimilarity search patterns, the binary and original value of the pixels are multiplied and converted with reference to the referenced pixel and its surrounding neighbours. The proposed segmentation produces an accuracy of 98.97% which is higher than any other existing algorithm. Depending upon the directions, the first-order derivatives are calculated for the projected information from the actual wave crested values. The efficiency of developed classification algorithm is found as 98.37% by effective combination with the horizontal edge density extraction. Finally, the proposed Inception Resnet V2 classification gives better accuracy than other segmentation method. Simulation results are included and performance analyses are tabulated for different weather conditions.
Keywords:
License plate recognition (LPR)
Bilateral filtering
Kernel search multiwavelet decomposition (KsMWD)
Inception Resnet V2 classification
Journal
IF:
3
Papers:
1.9W
Citations:
3.2W