稀疏表示 Resources

Showing items tagged with "稀疏表示"

Denoise images by applying sparse coding to local patches using pre-trained overcomplete dictionaries, followed by patch aggregation through averaging. This technique employs sparse and redundant representations over learned dictionaries, as detailed in "Image Denoising Via Sparse and Redundant Representations over Learned Dictionaries." The algorithm involves dictionary initialization, patch extraction, L1-norm optimization for sparse coding (e.g., via Orthogonal Matching Pursuit), and weighted averaging to reconstruct the denoised image.

MATLAB 214 views Tagged
Download

Compressive Sensing CS implementation featuring wavelet transform for sparse representation, Gaussian random matrix as measurement matrix, and IRLS algorithm for reconstruction. Processes 256x256 Lena image, compares original image with IRLS reconstruction results at different sampling ratios (0.74, 0.5, 0.3), runs 50 trials each to evaluate algorithm performance through PSNR metrics and execution time analysis.

MATLAB 232 views Tagged
Download

This toolbox implements the KSVD algorithm for adaptively achieving sparse signal representations with optimized sparsity characteristics through dictionary learning and sparse coding techniques.

MATLAB 270 views Tagged
Download

Implementation of a vehicle detection method using Haar features combined with SRC (Sparse Representation Classification). The provided files include training and test vehicle images. Note: The Haar features haven't been optimized due to time constraints, resulting in high dimensionality and slow sliding window processing. The code outputs performance statistics for reference, demonstrating sparse representation applications in vehicle detection. Key implementation aspects include feature extraction using Haar-like features and classification via sparse coding optimization algorithms.

MATLAB 239 views Tagged
Download