Generalised non‐locally centralised image de‐noising using sparse dictionary

Abstract

Recently, image de-noising algorithm based on sparse representation has received an increasing amount of attention. Such algorithms proposed a comprehensive sparse representation model, by solving the sparse coding problem and choosing the proper method for dictionary updating to achieve better de-noising results. Therefore, the construction of learning dictionary has become one of the key problems that limit the de-noising effectiveness. The non-locally centralised sparse representation de-noising algorithm uses principal component analysis method to achieve dictionary updating. Nevertheless, the instability of a single complete dictionary in sparse coding leads to erratic result in the process of the original image restoration. In this study, the authors present a new method named generalised non-locally centralised sparse representation algorithm. In the proposed method, the authors cluster the training patches extracted from a set of example images into subspaces, and then train dictionaries for subspaces by sparse analysis k-singular value decomposition dictionary, which is utilised to construct coded sub-block dictionary to avoid the instable results caused by a single dictionary. Experiments show that the improved method has better signal-to-noise ratio and de-noising effect compared with other methods.

Dialectical Analysis of Comparative Pedagogy Based on Multiple Intelligences Evaluation

“…Because of the strong specific problems, they do not have the rapidity, unity, and universality of the dialectic analysis scheme in comparative education; therefore, it is difficult to achieve the modular processing of dialectical intelligence based on the characteristics of comparative education data and comparative education [8]. Scholars found that the mainstream dialectic analysis system of comparative education has not developed to a very mature stage of technology, and more is based on specific problems or specific views of solution design [9]. Researchers found that there are still many shortcomings in the error rate of the comparative education model and the stability of relevant algorithms in the market.…”

Image reconstruction for emission tomography using intensity normalised patch‐based regularisation

“…For PET scanners, whose spatial resolution is significantly lower than that of X‐ray CT or MRI scanners, the CS‐based regularisation is usually limited to special applications where the pixel resolution needs to be increased to reconstruct high‐resolution images from significantly fewer measurements due to hardware limitations. Inspired by machine learning, which has been a topic in recent years, a variety of methods using dictionary learning have also been developed to solve the various image processing problems, such as denoising [18, 19] and super‐resolution [20]. While the dictionary learning methods have also been successfully applied in the medical imaging field [21], they usually require the additional data sets for the underlying images to train the dictionaries.…”

Image denoising using patch ordering and 3D transformation of patches

“…In recent years, in various image processing applications, much interest has been directed towards methods which utilise patches extracted from the image [13–16]. The extracted image patches are of a much smaller size compared to the overall image.…”

“…The main idea behind the non‐local methods is the utilisation of the similarity between patches which are placed in rather spatially distant locations of the image. Non‐local methods for image denoising such as non‐local means (NLM) filter [14, 17], non‐local principal component analysis [18, 19] and the recently proposed methods given in [15, 20] can outperform more traditional local approaches. A well‐known non‐local method is the NLM filter which obtains denoised estimate of a pixel by weighted averaging of pixels within similar patches placed at different locations of the image.…”