Ayyaz Hussain

Fuzzy based Impulse Noise Reduction Method

In this paper, we propose an image filtering technique based on fuzzy logic control to remove impulse noise for low as well as highly corrupted images. The proposed method is based on noise detection, noise removal and edge preservation modules. The main advantage of the proposed technique over the other filtering techniques is its superior noise removal as well as detail preserving capability. Based on the criteria of peak-signal-to-noise-ratio (PSNR), mean square error (MSE), structural similarity index measure (SSIM) and subjective evaluation measure we have found experimentally that the proposed method provides much better performance than the state-of-the-art filters. To analyze the detail preservation capability of the proposed filter sensitivity analysis is performed by changing the detail preservation module to see its effects on the details (texture and edge information) of resultant image. This sensitivity analysis proves experimentally that significant image details have been preserved by the proposed method.

Estimation and optimization based ill-posed inverse restoration using fuzzy logic

Intelligent systems ranging from neural network, evolutionary computations and swarm intelligence to fuzzy systems are extensively exploited by researchers to solve variety of problems. In this paper focus is on deblurring that is considered as an inverse problem. It becomes ill-posed when noise contaminates the blurry image. Hence the problem is very sensitive to small perturbation in data. Conventionally, smoothness constraints are considered as a remedy to cater the sensitivity of the problem. In this paper, fuzzy rule based regularization parameter estimation is proposed with quadratic functional smoothness constraint. For deblurring image in the presence of noise, a constrained least square error function is minimized by the steepest descent algorithm. Visual results and quantitative measurements show the efficiency and robustness of the proposed technique compared to the state of the art and recently proposed methods.

Quantum and impulse noise filtering from breast mammogram images

Recent advances in the field of image processing have shown that level of noise highly affect the quality and accuracy of classification when working with mammographic images. In this paper, we have proposed a method that consists of two major modules: noise detection and noise filtering. For detection purpose, neural network is used which effectively detect the noise from highly corrupted images. Pixel values of the window and some other features are used as feature for the training of neural network. For noise removal, three filters are used. The weighted average value of these three filters is filled on noisy pixels. The proposed technique has been tested on salt & pepper and quantum noise present in mammogram images. Peak signal to noise ratio (PSNR) and structural similarity index measure (SSIM) are used for comparison of proposed technique with different existing techniques. Experiments shows that proposed technique produce better results as compare to existing methods.

Fuzzy-based hybrid filter for Rician noise removal

Magnetic resonance images tend to be contaminated with random unwanted signals called noise, due to various reasons. Noise treatment of magnetic resonance brain images is considered as an important and challenging task for proper clinical and research investigations. In this manuscript, fuzzy logic-based hybrid Rician noise filter has been proposed. Proposed filtering technique uses estimated noise variance along with local and global statistics for the construction of a robust fuzzy membership function. Constructed fuzzy membership function assigns appropriate weights to the statistical estimates, based on their noise removal and detail preservation capability. Fuzzy weighted local and non-local estimators are then used for the restoration of a noisy pixel. Detailed simulations are performed, and restoration results are computed based on well-known performance measures. Numerical and visual results show that the proposed technique gives much better restored images than the existing methodologies.

Intensity-based statistical features for classification of lungs CT scan nodules using artificial intelligence techniques

A computer-aided diagnostic (CAD) system for effective and accurate pulmonary nodule detection is required to detect the nodules at early stage. This paper proposed a novel technique to detect and classify pulmonary nodules based on statistical features for intensity values using support vector machine (SVM). The significance of the proposed technique is, it uses the nodules features in 2D & 3D and also SVM for the classification that is good to classify the nodules extracted from the image. The lung volume is extracted from Lung CT using thresholding, background removal, hole-filling and contour correction of lung lobe. The candidate nodules are extracted and pruned using the rules based on ground truth of nodules. The statistical features for intensity values are extracted from candidate nodules. The nodule data are up-samples to reduce the biasness. The classifier SVM is trained using data samples. The efficiency of proposed CAD system is tested and evaluated using Lung Image Consortium Database (LIDC) that is standard data-set used in CAD Systems for Lungs Nodule classification. The results obtained from proposed CAD system are good as compare to previous CAD systems. The sensitivity of 96.31% is achieved in the proposed CAD system.

Color differences based fuzzy filter for extremely corrupted color images

Abstract In this paper, a color difference based fuzzy filter is presented for fix and random-valued impulse noise. Noise detection scheme of two stages was applied to detect noise efficiently whereas for noise removal an improved Histogram based Fuzzy Color Filter (HFC) is presented. Pixels detected as noisy by the noise detection scheme are deliberated as candidate for the removal of noise. Candidate noisy pixels are then processed using a modified Histogram based Fuzzy Color Filter to estimate their non-noisy values. The idea of using multiple fuzzy membership functions is presented, so that best suitable membership function for local image statistics can be used automatically. In the proposed technique we have used three different types of fuzzy membership functions (bell-shaped, trapezoidal-shaped, and triangular-shaped) and their fuzzy number construction algorithms are proposed. Experimentation is also performed with three, five, and seven membership functions. Type and number of suitable fuzzy membership functions are then identified to remove noise. Comparison with the existing filtering techniques is established on the basis of objective quantitative measures including structural similarity index measure (SSIM) and peak-signal-to-noise-ratio (PSNR). Simulations show that this filter is superior to that of the existing state-of-the-art filtering techniques in removing fix and random-valued impulse noise whereas retaining the details of the image contents.

Intelligent noise detection and filtering using neuro-fuzzy system

In this paper, we propose a neuro-fuzzy based blind image restoration to remove impulse noise from low as well as highly corrupted images. Main components of the proposed technique include noise detection, histogram estimation and noise filtering process. Proposed technique constructs the fuzzy sets using fuzzy number construction algorithm. These fuzzy sets are used in noise filtering process to remove impulse noise from the noisy pixels using neuro-fuzzy inference system and fuzzy decider. Experimental results are based on global and local error measures, which prove that the proposed technique gives superior results than the present well known impulse noise filtering methods.

Enhancing Fault Classification Accuracy of Ball Bearing Using Central Tendency Based Time Domain Features

Time-domain (TD) statistical features are frequently utilized in vibration-based pattern recognition (PR) models to identify faults in rotating machinery. Presence of possible fluctuations or spikes in random vibration signals can considerably affect the statistical values of the extracted features consequently. This paper discusses the sensitivity of TD features against the fluctuations occurred in vibration signals while classifying localized faults in ball bearing. Based on the sensitivity level, the features are statistically processed prior to employing a classifier in PR model. A central tendency-based feature pre-processing technique is proposed that enhances the diagnostic capability of classifiers by providing appropriate values. The feature processing reduces undesired impact of fluctuations on the diagnostic model. Several classifiers are utilized to evaluate the performance of the proposed method, and the results are evident of its effectiveness. The associated advantage of the feature pre-processing over the conventional pre-processing of raw data is its computational efficiency. It is worth mentioning that only few values in feature distributions are required to be processed rather than dealing with big TD vibration data set.

Reliable facial expression recognition for multi-scale images using weber local binary image based cosine transform features

Accurate recognition of facial expression is a challenging problem especially from multi-scale and multi orientation face images. In this article, we propose a novel technique called Weber Local Binary Image Cosine Transform (WLBI-CT). WLBI-CT extracts and integrates the frequency components of images obtained through Weber local descriptor and local binary descriptor. These frequency components help in accurate classification of various facial expressions in the challenging domain of multi-scale and multi-orientation facial images. Identification of significant feature set plays a vital role in the success of any facial expression recognition system. Effect of multiple feature sets with varying block sizes has been investigated using different multi-scale images taken from well-known JAFEE, MMI and CK+ datasets. Extensive experimentation has been performed to demonstrate that the proposed technique outperforms the contemporary techniques in terms of recognition rate and computational time.

Multi-stage binary patterns for facial expression recognition in real world

Facial expression recognition in real world has always been a challenging task in computer vision. Expressions are spawned from highly bendable features and vary in cultures, genders and ages among individuals. Moreover, complex backgrounds, intricate settings and varied illumination conditions add to complexity of recognition. Local binary patterns (LBP) capture the local region properties from images so these are extensively used for facial expression recognition. But still LBP and its variations including local gradient coding horizontal diagonal direction (LGC-HD), local gradient coding horizontal vertical diagonal direction (LGC-HVD) and compound local binary pattern (CLBP) are not able to resolve the issues of local illuminations that might result from disproportionate light, certain obscurities, shadows or use of accessories on the face. In proposed technique, multi-stage binary code is generated for each comparison against neighbouring pixel. LBP records only sign difference which eventually discards some significant texture information. In MSBP sign difference along with gradient difference accumulates changes along edges such as area around eye brows, eyes, mouth, bulges and wrinkles. MSBP is then compared with LBP, LGC-HD, LGC-HVD and CLBP. LGC-HD, LGCHVD and CLBP centered on experimentation using two approaches. These techniques are tested on dataset containing full face image named holistic and then division based approach. Experiments are performed using static facial expression in wild dataset. Results show that proposed approach MSBP outperforms other techniques with 96% accuracy in holistic approach and 60% accuracy in division based approach. Further, it is also ascertained from the experiments that all these approaches work better in holistic approach than division based approach.