Abstract:In this paper, we propose a novel image denoising algorithm exploiting features from both spatial as well as transformed domain. We implement intensity-invariance based improved grouping for collaborative support-agnostic sparse reconstruction. For collaboration firstly, we stack similar-structured patches via intensity-invariant correlation measure. The grouped patches collaborate to yield desirable sparse estimates for noise filtering. This is because similar patches share the same support in the transformed domain, such similar supports can be used as probabilities of active taps to refine the sparse estimates. This ultimately produces a very useful patch estimate thus increasing the quality of recovered image by discarding the noise-causing components. A region growing based spatially developed post-processor is then applied to further enhance the smooth regions by extracting the spatial domain features. We also extend our proposed method for denoising of color images. Comparison results with the state-of-the-art algorithms in terms of peak signal-to-noise ratio (PNSR) and structural similarity (SSIM) index from extensive experimentations via a broad range of scenarios demonstrate the superiority of our proposed algorithm.
What problem does this paper attempt to address?
This paper attempts to solve the key problems in image denoising, especially the challenges when dealing with high - noise, different image sizes, and natural or man - made images. Specifically, the paper proposes a new image denoising algorithm, aiming to improve the denoising effect by utilizing the features of the spatial domain and the transform domain while preserving the details of the image. The following are the main contributions of the paper:
1. **Problem Background**:
- Image denoising is of great significance in fields such as statistical signal processing, seismic signal processing, computer vision, the Internet of Things (IoT), robotics, and artificial intelligence.
- Noise pollution will affect the reliability of image analysis, especially during the image acquisition stage.
- Image denoising is an ill - posed problem, and it is theoretically difficult to accurately recover the original image from the noise.
2. **Limitations of Existing Methods**:
- Existing denoising methods are mainly divided into spatial - domain - based methods and transform - domain - based methods.
- Spatial - domain methods such as total variation minimization and non - local means (NL - means) algorithm perform well in smooth region and edge preservation.
- Transform - domain methods such as wavelet transform and sparse representation are effective in removing noise, but have high computational complexity.
3. **Contributions of the Paper**:
- A new image denoising algorithm, called Collaborative Dual - Domain Patch Filtering (C2DF), is proposed.
- The algorithm is implemented through the following steps:
1. **Image Decomposition**: Decompose the image into overlapping blocks and find structurally similar blocks.
2. **Sparse Estimation**: Perform collaborative denoising in the sparse domain and use the Bayesian method to recover the sparse representation.
3. **Post - processing**: Apply spatial - domain post - processing based on region growing to further enhance the smooth regions.
4. **Specific Methods**:
- **Intensity Invariant Grouping**: Use the correlation coefficient instead of the Euclidean distance to find structurally similar blocks, thereby increasing the number of collaborative blocks.
- **Sparse - Domain Collaboration**: Improve the accuracy of sparse estimation by calculating the probability of active positions.
- **Dictionary Decorrelation**: Reduce the correlation between dictionary columns to improve the robustness of calculation and performance.
- **Image Reconstruction**: Reconstruct the denoised image by averaging the results of overlapping blocks.
5. **Experimental Results**:
- Compared with existing advanced algorithms (such as NL - means, K - SVD, and BM3D), C2DF performs excellently in terms of peak signal - to - noise ratio (PSNR) and structural similarity index (SSIM).
- Especially when dealing with high - noise and different - sized images, C2DF can better preserve the details of the image, reducing blurring and artifacts.
In conclusion, this paper proposes an innovative image denoising algorithm, which significantly improves the denoising effect by combining the features of the spatial domain and the transform domain, especially when dealing with high - noise and complex images.