Multiscale vehicle image enhancement based on hybrid chaotic particle swarm algorithm

Image enhancement is essential for applications like automatic license plate recognition, where accuracy is crucial for extracting information. This paper presents a gray-scale image enhancement technique that combines chaotic particle swarm optimization (CPSO) with simulated annealing (SA) at multiple image scales. The process starts with Laplace pyramid decomposition to generate multi-level images with varying resolutions. The CPSO and SA algorithms are then applied to optimize parameters, enhancing image details and contrast at each scale. Afterward, the enhanced layers are reconstructed into a final image. Simulation results show that this method outperforms traditional image enhancement techniques in terms of visual quality, clarity, and detail, which are critical for license plate recognition.