China 
Africa 
Explore chapters and articles related to this topic
A Survey of Artificial Intelligence Techniques Used in Biometric Systems
Published in Chiranji Lal Chowdhary, Intelligent Systems, 2019
C. N. Patill, E. Naresh, B. P. Vijay Kumar, Prashanth Kambli
In facial recognition biometric system, face image of the user is captured, and Haar-like feature is used to recognize the object in the real-time captured image. Haar-based cascade classifiers are used to distinguish between various types of Haar features, like Edge feature, Line feature, four rectangle feature, and center-surrounded features. Haar features are provided as a contribution to the course classifiers (Fig. 2.3). This Haar feature can be calculated by assuming the contrast between aggregate of pixels in white rectangular part or section and aggregate of pixels in dark rectangular area at various aspects of orientations, proportions, and scales.8
Driver Drowsiness Detection using Image Processing Technique
Published in Purna Chandra Mishra, Muhamad Mat Noor, Anh Tuan Hoang, Advances in Mechanical and Industrial Engineering, 2022
R. Seetharaman, C. Kaushik Viknesh, A. Murugan, S. Viswanathan, G. K. Pandiyarajhan
Viola Jones Algorithm forms the basis of this robust system. Viola Jones Algorithm helps in achieving the high detection rates. It also processes the images rapidly. It forms the basis of most of the real time systems, as it works only on the present single grey scale image. There are three main parts in this algorithm: Integral Image which allows very fast feature evaluation.Classifier function which is built using small number of important features.Method of combining the classifiers in a cascade structure to increase the speed of the detector by focusing on the promising regions of interest. Viola Jones Algorithm is fast, efficient and gives level of accuracy. First step of Viola Jones Algorithm includes training of the dataset, using any machine learning algorithm. The training is done using the two sets of positive and negative images. Basically, this training is done in order to develop predictive relationship between the datasets. In the Viola–Jones object detection framework, the Haar-like features are therefore organised in something called a classifier cascade, to form a strong learner or classifier. The key advantage of a Haar-like feature over most other features is its calculation speed. A Haar-like feature of any size can be calculated in constant time (approximately 60 microprocessor instructions for a 2- rectangle feature). After the classifier function is developed, AdaBoost algorithm is used for selecting the required features and training the dataset. This algorithm basically is used for enhancing the performance of the classifier function. AdaBoost is an effective procedure for searching out a small number of good “features” which nevertheless have significant variety. This algorithm is used for selecting the features like eyes or mouth region [6]. After selecting the features of interest, the datasets are trained.
A Technique for Human Upper Body Parts Movement Tracking
Published in IETE Journal of Research, 2022
Krishan Kumar, Abhya Mishra, Sanjay Dahiya, Ajay Kumar
A Haar-like feature consists of particular rectangular spaces at a different specific location for the discovery of the window; the process needs summing the pixel intensities in each region and after that calculates the difference between those sums [16]. This variation is used to maintain the subsections of a detecting picture. For illustration, consider that there is an image corpus of human faces. One common perception with all faces is that the eye area is more sinister than the cheeks. Consequently, a primary Haar characteristic for face discovery is placed within two neighboring intersections beyond the eye and face area. The position of these squares is defined as corresponding to a discovery window. It acts as a bounding box to the target object. In the discovery phase of the Viola-Jones object detection framework, a window of target size is moved over the input image. For each subsection of the image, the Haar-like feature is estimated. This difference is compared to a learned threshold that separates non-objects from objects. Because such a Haar-like feature is only a weak learner or classifier, many Haar-like features are necessary to describe an object with sufficient accuracy. The Haar-like features are organized in a classifier cascade in the VJA object detection framework to form a strong learner or classifier.
Automatic segmentation of the thumb trapeziometacarpal joint using parametric statistical shape modelling and random forest regression voting
Published in Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, 2019
Marco T. Y. Schneider, Ju Zhang, Joseph J. Crisco, Arnold-Peter C. Weiss, Amy L. Ladd, Poul M. F. Nielsen, Thor Besier
Haar-like features are used in image object recognition and have been used in image segmentation in 2D (Cootes et al. 2012) and 3D (Norajitra et al. 2015; Norajitra and Maier-Hein 2017). Haar-like features are calculated by comparing the difference in summed pixel intensity between regions of pixels (in 2D) or voxels (in 3D) within a bounding box These regions may be labelled as ‘dark’ and ‘light’, where the bounds of the ‘light’ region can be randomised to create an infinite set of 3D Haar-like features (Lindner et al. 2013; Norajitra and Maier-Hein 2017). Due to this formulation, 3D Haar-like features do not support more complex features that have more than one ‘light’ region, such as chequered 3D Haar-like features, which can provide important textural information. In this study, we used a fixed set of eight 3D Haar-like features (Figure 2), including three features with one ‘light’ region, three features with two ‘light’ regions, three features with axis aligned chequered regions (two ‘light’ and two ‘dark’ regions), and one feature with completely chequered regions (four ‘dark’ and four ‘light’ regions). These features can be calculated efficiently by precomputing the integral image using Equation 2. The difference in summed pixel intensity can then be calculated using Equation 3 (Norajitra and Maier-Hein 2017).
Vertebral corners detection on sagittal X-rays based on shape modelling, random forest classifiers and dedicated visual features
Published in Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, 2019
Shahin Ebrahimi, Laurent Gajny, Wafa Skalli, Elsa Angelini
Variability of manual landmark localisation at each vertebral level was evaluated in (Champain et al. 2006) but on shape parameters different than the ones evaluated in our study. However, it seems that our orientation errors are slightly higher than manual uncertainty in asymptomatic subjects, while in pathologic cases they are similar. A major contribution of our proposed methodology is the design of our pool of image features, composed of three complementary types. Although the number of Haar-like and HOG features is smaller than the CF ones, all three types contribute as predictive features for the RF classifiers. HOG features have proven their efficiency in encoding robust local visual characteristics. The tailored Haar-like features provide specific discriminative power for corner detection and determining the corner types. Feature importance analysis of our proposed and tested Haar-like features shows that the features in Equation (5) are among the strongest predictive features (best descriptors being O4 and O5). This category of Haar-like features is novel and was not exploited in previously published papers. However, in general the spatial range of Haar-like features is limited. Adding some contextual features lets us explore large neighbouring regions around the candidate points, which leads to a reinforced discrimination against false positives. Reported results show that the combination of these three feature types improves the precision of the detected corner positions. Regarding contextual features (CF), while various types have been developed and used in previous studies, our study presents an original design tuned for the given detection task of precise vertebrae corner localisation and identification on X-ray images