Multi-scale vehicle image enhancement based on hybrid chaotic particle swarm algorithm

Image enhancement plays a crucial role in the process of image recognition, especially in applications such as automatic license plate recognition, where clarity and accuracy are essential for extracting precise information from images. In order to address the challenge of improving the recognition quality of license plates, this paper introduces an advanced gray-scale image enhancement technique. This method integrates the chaotic particle swarm optimization (CPSO) algorithm with the simulated annealing (SA) algorithm at different image scales, effectively optimizing the enhancement process.The method begins by decomposing the original image using the Laplace pyramid decomposition, which generates a series of multi-level images, each containing information from different scales or resolutions. By doing so, we can isolate and enhance specific image details more effectively at each level. Next, the chaotic particle swarm and simulated annealing algorithms are employed, leveraging their respective strengths in global search and local optimization. Specifically, the particle swarm algorithm provides a mechanism for exploring the parameter space, while the simulated annealing algorithm refines the solutions by preventing premature convergence. A hybrid perturbation operator is applied to the local optimal solution at each scale to further enhance the image's details and contrast.Finally, all the enhanced layers are reconstructed back into a single image, thereby completing the image enhancement process. Extensive simulation experiments were conducted, comparing this method with other traditional image enhancement algorithms. The experimental results demonstrate that the proposed technique yields superior visual quality, effectively improving image clarity and detail, which is crucial for tasks such as license plate recognition.