China 
Africa 
Explore chapters and articles related to this topic
Continuous-Wave Doppler Radar for Human Gait Classification
Published in Moeness G. Amin, Radar for Indoor Monitoring, 2017
Fok Hing Chi Tivive, Abdesselam Bouzerdoum, Bijan G. Mobasseri
The contrast-enhanced patches are convolved with a set of log-Gabor filters to extract discriminative features, such as motion energy at different spatial frequencies (or scales) and orientations. Compared to their traditional Gabor counterparts, log-Gabor filters have neither direct current component nor bandwidth limitation [55]. Therefore, a small set of filters is sufficient to cover the desired frequency spectrum. In the frequency domain, a log-Gabor filter is given by () Hk,l(f,θ)=exp{−[log(ffk)]22[log(β)]2}exp{−(θ−θl)22σθ2}
Evaluation
Published in Francesco Banterle, Alessandro Artusi, Kurt Debattista, Alan Chalmers, Advanced High Dynamic Range Imaging, 2017
Francesco Banterle, Alessandro Artusi, Kurt Debattista, Alan Chalmers
The subband error signal stage generates a spatial error between the reference and distorted signal. log-Gabor filters are employed to find differences at distinct scales and orientations. The subbands are computed via inverse DFT of the product of the signal with the log-Gabor filter. The error is then given as Errt,s,0=2lt,s,0srclt,s,0dis+ε(lt,s,0src)2+(lt,s,0dis)2+ε, $$ Err_{{t,s,0}} = \frac{{2l_{{t,s,0}}^{{src}} l_{{t,s,0}}^{{dis}} + \varepsilon }}{{(l_{{t,s,0}}^{{src}} )^{2} + (l_{{t,s,0}}^{{dis}} )^{2} + \varepsilon }}, $$
Review on Pupil Segmentation Using CNN-Region of Interest
Published in Kamal Kumar Sharma, Akhil Gupta, Bandana Sharma, Suman Lata Tripathi, Intelligent Communication and Automation Systems, 2021
A. Swathi, Aarti, Sandeep Kumar
The authors in Reference [12] performed feature extraction by using a reduced-pixel block algorithm on CASIA version 3, with 91% accuracy in matching. They used edge detection and Hough transform algorithms to draw the iris and pupil. Local binary patterns were used to represent texture patterns which in turn used the Log-Gabor filter. Further they extended it to other features by supplying this as input to SVM.
Feature level fusion framework for multimodal biometric system based on CCA with SVM classifier and cosine similarity measure
Published in Australian Journal of Electrical and Electronics Engineering, 2023
Chetana Kamlaskar, Aditya Abhyankar
In this work, a feature vector of fixed length is created by extracting features from preprocessed images. The IrisCode (Daugman 2004) is formed using Log-Gabor filter while the FingerCode (Jain et al. 2000) is formed using a bank of Gabor Filter. For images of iris, normalisation is carried out after segmentation in order to reduce the impact of iris image scale changes. For this, the iris region is mapped in 20(r) x 240(θ) fixed dimensions of polar coordinates which indicates radial (r) and angular (θ) resolution of the normalised image. The selection of these values is a compromise between noise removal and the acquisition of reasonable size templates (Daugman 2004). In the next step of feature extraction, normalised iris images are convolved with a log-Gabor filter and then encoded in a binary vector with a fixed 9600 × 1 length as described in (Kamlaskar and Abhyankar 2020, 2021).
A novel convolutional neural network with gated recurrent unit for automated speech emotion recognition and classification
Published in Journal of Control and Decision, 2023
P. Ravi Prakash, D. Anuradha, Javid Iqbal, Mohammad Gouse Galety, Ruby Singh, S. Neelakandan
The feature extraction and feature selection can improve the learning efficiency, by reducing computation complexity, developing higher generalised methods, and reducing needed memory. The final phase of speech emotion identification is classification. It includes categorising the raw data in the creation of utterance or frame utterance into a specific type of emotion which is based on the feature extraction from the information. Recently, in speech emotion identification, scientists presented several classifying techniques, like the Gaussian mixture method (GMM), hidden Markov method (HMM), support vector machine (SVM), neural networks (NN), and recurrent neural network (RNN). Few kinds of classification are also presented by several scientists like an altered brain emotional learning method (BEL) where the adjustive neuro-fuzzy inference system (ANFIS) and multilayer perceptron (MLP) are combined for speech emotion detection. The Voiced Segment Selection (VSS) technique is presented in (Zehra et al., 2021) copes with the voiced signal segment as a texture image processing feature that is distinct from the conventional technique. It utilises the Log-Gabor filter for extracting the voice and unvoiced feature from the spectrogram to create the class.
A Vehicle License Plate Detection and Recognition Method Using Log Gabor Features and Convolutional Neural Networks
Published in Cybernetics and Systems, 2023
Ahmed Zaafouri, Mounir Sayadi, Wei Wu
Where the radial component of the logarithmic log Gabor filter is given by: is the filter’s center frequency and controls the scale bandwidth of the filter and the orientation of the filter is defined by: define the angular bandwidth of the filter and represents the orientation angle of the filter. Figure 3 shows the 3 D shape of log-Gabor filter in the frequency domain.