Multi-Layer License Plate Recognition by using Frame Grabbing & Neural Network
Prof. Priyanka Kumbhar1, Sakshi Harihar2, Swapnali Dudhade3, Yadneysh Raut4
1 Prof. Priyanka Kumbhar, Information Technology, P.G. Moze college of engineering
2 Sakshi Harihar, Information Technology, P.G. Moze college of engineering
3 Swapnali Dudhade, Information Technology, P.G. Moze college of engineering
4 Yadneysh Raut, Information Technology, P.G. Moze college of engineering
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - This project introduces a sophisticated system for License Plate Recognition (LPR) by integrating frame grabbing techniques with neural network architecture. The primary objective is to develop an efficient and accurate method for real-time license plate identification from video streams. Frame grabbing is employed as the initial step to extract frames from video inputs, enabling precise segmentation of license plate regions. These frames undergo preprocessing to enhance image quality and standardize features, ensuring optimal input for the subsequent neural network stages.
The neural network constitutes a multi-layered structure, designed to process the extracted frames and perform intricate pattern recognition tasks. Leveraging advanced algorithms and deep learning principles, this network is trained on diverse datasets encompassing various license plate designs, sizes, fonts, and environmental conditions. The model's adaptability is honed through extensive training to enable robust feature extraction and character recognition.
The system's performance is rigorously evaluated using benchmark datasets and real-world video sequences, measuring accuracy, speed, and reliability in license plate detection and character recognition. Experimental results showcase the system's capability to accurately identify license plates in challenging scenarios, including varying lighting conditions, plate orientations, and distances from the camera.
The proposed approach demonstrates significant potential for practical implementation in surveillance, traffic management, and law enforcement applications. Its ability to swiftly and accurately process video streams in real time offers promising prospects for enhancing security and efficiency in diverse operational contexts.
Key Words: LPRS, Character Segmentation, Character Recognition, Back Propagation Neural Network
