9 July 2008 in IEEE Transactions on Industrial Electronics
This paper presents a new approach for the visual localization, detection, and classification of various nonstationary power signals using a variety of windowing techniques. Among the various windows used earlier like sine, cosine, tangent, hyperbolic tangent, Gaussian, bi-Gaussian, and complex, the modified Gaussian window is found to provide excellent normalized frequency contours of the power signal disturbances suitable for accurate detection, localization, and classification. Various nonstationary power signals are processed through the generalized S -transform with modified Gaussian window to generate time-frequency contours for extracting relevant features for pattern classification. The extracted features are clustered using fuzzy C-means algorithm, and finally, the algorithm is extended using either particle swarm optimization or genetic algorithm to refine the cluster centers.
22 February 2013 in IEEE Transactions on Industrial Electronics
This paper proposes an empirical-mode decomposition (EMD) and Hilbert transform (HT)-based method for the classification of power quality (PQ) events. Nonstationary power signal disturbance waveforms are considered as the superimposition of various undulating modes, and EMD is used to separate out these intrinsic modes known as intrinsic mode functions (IMFs). The HT is applied on all the IMFs to extract instantaneous amplitude and frequency components. This time-frequency analysis results in the clear visual detection, localization, and classification of the different power signal disturbances. The required feature vectors are extracted from the time-frequency distribution to perform the classification. A balanced neural tree is constructed to classify the power signal patterns. Finally, the proposed method is compared with an S-transform-based classifier to show the efficacy of the proposed technique in classifying the PQ disturbances.
1 June 2010 in Applied Soft Computing
10 January 2018 in IET Image Processing
Several diseases can be detected and diagnosed at a very early stage by observing changes in retinal blood vessel features. In order to detect these changes, the major step required is blood vessel segmentation. In this study, an effective method for the segmentation of blood vessels on coloured digital retinal images has been proposed. The proposed method uses linear combination of line detectors at varying scales along with multiple windows of different sizes. By implementing this technique, the drawbacks encountered in multi‐scale line detection such as noise and false vessel detection around the optic disk are eliminated. The performance of the proposed method is evaluated on three publicly available datasets DRIVE, STARE and CHASE by considering sensitivity, specificity, accuracy, precision, false discovery rate, F1 score, Matthews correlation coefficient and G‐mean. The results were analysed by comparing with most of the other earlier existing methods and have proven to achieve higher accuracy. It has also confirmed its effectiveness and robustness with high‐resolution retinal images together with better simplicity and faster implementation for reliable blood vessel segmentation.
1 January 2009 in Applied Soft Computing
1 June 2012 in Swarm and Evolutionary Computation