Explore MATLAB source code curated for "改进" with clean implementations, documentation, and examples.
An improved mathematical morphology-based small target detection method for infrared imagery, delivering higher detection accuracy than conventional Top-Hat transformation approaches through optimized structuring elements and multi-scale analysis.
Enhanced global particle swarm optimization algorithm incorporating convergence factors and adaptive inertia weight from contemporary research, demonstrating superior efficiency compared to traditional basic implementations
My improvement on the TVL1 method for removing facial illumination to obtain facial texture images, which significantly enhances face recognition under varying lighting conditions. The approach also supports facial feature localization through optimized gradient-based processing and illumination decomposition algorithms.
L2-norm-based multi-frame image super-resolution reconstruction algorithm improves computational accuracy and efficiency over existing approaches, implemented through optimized matrix operations and regularization techniques.
The improved nonlinear anisotropic diffusion method for edge-preserving image denoising incorporates second-order derivatives to achieve superior filtering performance and better edge preservation.
A newly proposed swarm intelligence algorithm called Chicken Swarm Optimization (CSO), available for download and improvement with implementation-specific enhancements.
Implementation of digital watermarking using Discrete Cosine Transform (DCT) - seeking suggestions for code optimization and algorithm improvements
This implementation presents an enhanced compressed sensing signal recovery algorithm that improves upon the greedy iterative Orthogonal Matching Pursuit (OMP) method. The conventional OMP algorithm selects suboptimal atoms during each iteration, failing to maximize residual reduction. Our Optimised_OMP algorithm ensures selected atoms remain orthogonal to the subspace spanned by previously chosen atoms, enabling faster residual reduction and accelerated convergence. The code implements optimal atom selection through Gram-Schmidt orthogonalization or QR decomposition techniques.
Real-time motion foreground detection using Gaussian Mixture Model (GMM) with improved performance over traditional single Gaussian - features enhanced segmentation and adaptability to dynamic scenes
Implementation of color transfer between images using Reinhard's algorithm with Reinhard.m program, featuring an improved version Reinhard_improved.m with enhanced color mapping and computational efficiency