Noraziahtulhidayu Kamarudin

Denoising of natural images through robust wavelet thresholding and genetic programming

Digital images play an essential role in analysis tasks that can be applied in various knowledge domains, including medicine, meteorology, geology, and biology. Such images can be degraded by noise during the process of acquisition, transmission, storage, or compression. The use of local filters in image restoration may generate artifacts when these filters are not well adapted to the image content as a result of the heuristic optimization of local filters. Denoising methods based on learning procedure are more capable than parametric filters for addressing the conflicts between noise suppression and artifact reduction. In this study, we present a nonlinear filtering method based on a two-step switching scheme to remove both salt-and-pepper and additive white Gaussian noises. In the switching scheme, two cascaded detectors are used to detect noise, and two corresponding estimators are employed to effectively and efficiently filter the noise in an image. In the process of training, a method according to patch clustering is utilized, and genetic programming (GP) is subsequently applied to determine the optimum filter (wavelet-domain filter) for each individual cluster, while in testing part, the optimum filter trained beforehand by GP is recovered and used on the inputted corrupted patch. This adaptive structure is employed to cope with several noise types. Experimental and comparative analysis results show that the denoising performance of the proposed method is superior to that of existing denoising methods as per both quantitative and qualitative assessments.

Natural image noise level estimation based on local statistics for blind noise reduction

This study proposes an automatic noise estimation method based on local statistics for additive white Gaussian noise. Noise estimation is an important process in digital imaging systems. For example, the performance of an image denoising algorithm can be significantly degraded because of poor noise level estimation. Most of the literature on the subject tends to use the true noise level of a noisy image when suppressing noise artifacts. Moreover, even with the given true noise level, these denoising techniques still cannot attain the best result, particularly for images with complicated details. In this study, a patch-based estimation technique is used to estimate for noise level and applies it to the proposed blind image denoising algorithm. Our approach includes selecting low-rank sub-image with removing high-frequency components from the contaminated image. This selection is according to the gradients of patches with the same statistics. Consequently, we need to estimate the noise level from the selected patches using principal component analysis (PCA). For blind denoising applications, the proposed denoising algorithm integrates the undecimated wavelet-based denoising algorithms and PCA to develop the subjective and objective qualities of the observed image, which result from filtering processes. Experiment results depict that the suggested algorithm performs efficiently over a wide range of visual contents and noise conditions, as well as in additive noise. Associated with different conventional noise estimators, the proposed algorithm yields the best performance, higher-quality images, and faster running speed.

Feature extraction using Spectral Centroid and Mel Frequency Cepstral Coefficient for Quranic Accent Automatic Identification

This paper presents the process of Quranic Accent Automatic Identification. Recent feature extraction technique that is used for Quranic verse rule identification/Tajweed include Mel Frequency Cepstral Coefficients (MFCC) which prone to additive noise and may reduce the classification result. Therefore, to improve the performance of MFCC with addition of Spectral Centroid features and is proposed for used in feature extraction of Quranic accents. Through implementing the Spectral Centroid Feature, it complements in improving the accuracy result of identifying the Quranic accents. The pattern classification algorithm here used the dimensional reduced technique from Probabilistic Principal Component Analysis (PPCA) on the features and Gaussian Mixture Model, in purpose to model the effectiveness of both combination of feature extraction. The accuracy of automatic identification for such Quranic Accents are found increasing from 96.9% to 100% with the application of SCF.

Natural image noise removal using non local means and hidden Markov models in stationary wavelet transform domain

In self-similarity digital image features, nonlocal means (NLM) exploits the major aspects when it comes to noise removal methods. Despite the high performance characteristics that NLM has proven, computational complexity yet to be highly achieved especially in case of complicated texture patches. In this regard, this study uses the clustered batches of noisy images and hidden Markov models (HMMs) in order to achieve noiseless images where the dependency between additive noise model pixels and its neighbors on stationary wavelet transform is found using HMMs. This paper is helpful and significant in order to develop a speedy and efficient plant recognition system computer-based to identify the plant species. The pivotal significant of the use of NLM and HMMs in this study is to ensure the statistical properties of the wavelet transform such as multiscale dependency among the wavelet coefficients, local correlation in neighbourhood coefficients. Practically, the experimental results present that the proposed algorithm has depicts high visual quality images in the experiments that are conducted in this study, apart from the objective analysis of the proposed algorithm, the execution time and its complexity show a competitive performance with state of the art noise removal methods in low and high noise levels.

Natural image noise removal using nonlocal means and hidden Markov models in transform domain

Nonlocal means (NLM) which utilizes the self-similarity is considered as one of the most popular denoising techniques. Although NLM can attain significant performance, it shows a few loopholes, such as its computational complexity when it comes to similarity measurements, and the small number of sufficient candidates that use to choose the target patches which have complicated textures. In this paper, the use of clustering based on moment invariants and the hidden Markov model (HMM) is proposed to achieve preclassification and thus capture the dependency between additive white Gaussian noise pixel and its neighbors on the wavelet transform. The HMM also allows hidden states to connect to one another to capture the dependencies among coefficients in the transform domain. In the practical part, the experimental results present that the proposed algorithm has the ability to show denoised images better than the results of state-of-the-art denoising methods both objectively in peak signal-to-noise ratio and structural similarity and subjectively using visual results, especially when the noise level is high.

Sequential parameterizing affine projection (SPAP) windowing length for acoustic echo cancellation on speech accents identification

Echo cancellation has always in the preprocessing steps before the signals are converted to feature vectors and pattern classification. This is always the correct flow of speech identification. Therefore, in order to get the best cleaned signal, the usage of adaptive echo cancellation removed the echo and also the noise which deteriorates the signals and final results during classification process. The concepts of windowing length may improve the cleaned signals acquired after the noise or echo cancellation process is done. By proposing the preconfigured windowing length through sequential technique, the results is giving improvement from normal length of 200ms to 400ms whereby the results of Word Error Rate(WER), Equal Error Rate (EER) and accuracies can be viewed with increases around 5–10% of percentage values compared with echoed signal and reduced the WER and EER too with applying of the sequential parameterization (SPAP) technique.

Al-quran learning using mobile speech recognition: An overview

The usage of mobile application in various aspects has been worldwide accepted and there are variety of mobile applications which developed to cater the usage of different background of the user. In this paper, a short survey which includes questionnaire is distributed to find the interest of user whom using application for learning Quran and concept of mobile speech apps. The main interest of this survey is to find the acceptance of user and explanation on the proposed usage of mobile speech recognition with feature of learning apps. Factors of mobile speech recognition and mobile learning are listed to support the results from the short survey.