Swati Nigam

Multiresolution approach for multiple human detection using moments and local binary patterns

Human detection is a central problem in development of any surveillance application. In this study, we present a simple and efficient, multi-resolution gray scale invariant approach for multiple human detection. The multiresolution is important for objects of different size and gray scale invariance is important due to uneven illumination and within-class variability. The proposed method is based on integration of central moments upon multi-resolution gray scale invariant local binary patterns operator. Since, the local binary patterns operator is invariant against different resolutions of space scale and monotonic change in gray scale, therefore the proposed method is robust in terms of variations in space scale as well as gray scale. Another advantage is high computational accuracy of the method due to use of moment operator which enhances the efficiency of the proposed method. Moreover, the proposed method is simple, as these operations can be performed within a few steps in a small neighborhood and a lookup table. The proposed method is tested on multiple human images and experimentally found appropriate for multiple human detection. The proposed method has been evaluated over two datasets, one is our own created dataset and the other is standard INRIA human detection dataset. Experimental results obtained from the proposed method demonstrate that better discrimination can be achieved for human and non-human objects in real scenes.

An effective local feature descriptor for object detection in real scenes

In this study, we advocate the importance of robust local features that allow object form to be distinguished from other objects for detection purpose. We start from the grid of Histogram of oriented gradients (HOG) and integrate Scale Invariant Feature Transform (SIFT) within them. In HOG features an objects appearance is detected by the distribution of local intensity gradients or edge directions for different cells. In the proposed method we have computed the SIFT despite of computing intensity gradients for these cells. In this way, the proposed approach does not only provide more significant information than just providing intensity gradients but also proves to deal with following challenges: (i) scale invariance; (ii) rotation invariance; (iii) change in illumination; and (iv) change in view points. With qualitative and quantitative experimental evaluation on standard INRIA dataset, we have compared the proposed method with other state of the art object detection methods and demonstrated better performance over them.

Moment invariants based object recognition for different pose and appearances in real scenes

Object recognition in real scenes is a central problem in computer vision. In this paper we propose a new approach for shape based recognition of objects in real scenes. This approach uses moment invariants for identification of shape features. Moment Invariants are functions of central moments. They are invariant against linear transformations such as rotation, translation and scaling. Therefore, their integration provides recognition of objects in real scenes with different pose and appearances. In this way, the proposed approach does not only provide invariant object recognition, but also capable of dealing with challenges like variation in pose and appearances. We have used linear support vector machine (SVM) for classification of object and non-object data. With qualitative and quantitative experimental evaluation on standard INRIA Pedestrian dataset, we have compared performance of the proposed method with other state of the art shape feature descriptors based object recognition methods and demonstrated better performance over them.

Multifont Oriya Character Recognition Using Curvelet Transform

In this paper, we have proposed a new character recognition method for Oriya script which is based on curvelet transform. Multi font Oriya character recognition has not been attempted previously. Ten popular Oriya fonts have been used for the purpose of character recognition. The wavelet transform has widely been used for character recognition purpose, but it cannot well describe curve discontinuities. We have used curvelet transform for recognition which is done using curvelet coefficients. This method is suitable for Oriya character recognition as well as various other scripts’ recognition purpose also. The proposed method is simple and extracts effectively the features in target region, which characterizes better and represents more robustly the characters. The experimental results validate that the proposed method improves greatly the recognition accuracy and efficiency than other traditional methods.

Automatic human activity recognition in video using background modeling and spatio-temporal template matching based technique

Human activity recognition is a challenging area of research because of its various potential applications in visual surveillance. A spatio-temporal template matching based approach for activity recognition is proposed in this paper. We model the background in a scene using a simple statistical model and extract the foreground objects in a scene. Spatio-temporal templates are constructed using the motion history images (MHI) and object shape information for different human activities in a video like walking, standing, bending, sleeping and jumping. Experimental results show that the method can recognize these multiple activities for multiple objects with accuracy and speed.

Integration of moment invariants and uniform local binary patterns for human activity recognition in video sequences

In this study, we present a method for human activity recognition in video sequences. Human activities are often described by a holistic feature vector comprising of a set of local motion descriptors. Here, we use a novel local shape feature descriptor for human activity recognition which is an integration of moment invariants and uniform local binary patterns (MI_ULBP). This feature descriptor is passed to a binary support vector machine pattern classifier for classification of human activities. Activity recognition is achieved through probabilistic search of image feature database representing previously seen activities. Experiments are performed over four benchmark video datasets Weizmann, KTH, CASIA and Collective human activity. Visual results and quantitative comparisons with existing methods show that the proposed method gives better recognition of human activities in video sequences with varying backgrounds and viewpoints.

On human activity recognition in video sequences

In this paper, we describe a novel template matching based approach for recognition of different human activities in a video sequence. We model the background in the scene using a simple statistical model and extract the foreground objects present in a scene. The matching templates are constructed using the motion history images (MHI) and spatial silhouettes for recognizing activities like walking, standing, bending, sleeping and jogging in a video sequence. Experimental results demonstrate that the proposed method can recognize these activities accurately for standard KTH database as well as for our own database.

Curvelet transform based object tracking

In this paper, we have proposed a new object tracking method in video sequences which is based on curvelet transform. The wavelet transform has widely been used for object tracking purpose, but it cannot well describe curve discontinuities. We have used curvelet transform for tracking. Tracking is done using energy of curvelet coefficients in sequence of frames. This method is suitable for object tracking as well as human object tracking purpose also. The proposed method is simple and does not require any other parameter except curvelet coefficients. Experimental results demonstrate performance of this method.

Multiscale Local Binary Patterns for Facial Expression-Based Human Emotion Recognition

Facial expression is an important cue for emotion recognition in human behavior analysis. In this work, we have improved the recognition accuracy of facial expression recognition and presented a system framework. The framework consists of three modules: image processing, facial features extraction, and facial expression recognition. The face preprocessing component is implemented by cropping the facial area from images. The detected face is downsampled by bilinear interpolation to reduce the feature extraction area and to enhance execution time. For extraction of local motion-based facial features, we have used rotation-invariant uniform local binary patterns (LBP). A hierarchical multiscale approach has been adopted for computation of LBP. The selected features were fed into a well-designed tree-based multiclass SVM classifier with one-versus-all architecture. The system is trained and tested with benchmark dataset from JAFFE facial expression database. The experimental results of the proposed techniques are effective toward facial expression recognition and outperform other methods.

Contourlet transform based moving object segmentation

Moving object segmentation is an important step toward development of any computer vision systems. In the present work, we have proposed a new method for segmentation of moving objects, which is based on single change detection method applied on Contourlet coefficients of two consecutive frames. We have chosen contourlet transform as it has high directionality and represents salient features of image such as edges, curves and contours in better way as compared with wavelet transform. The proposed method is simple and does not require any other parameter except contourlet coefficients. Results after applying the proposed method for segmentation of moving objects are compared with other state-of-the-art methods in terms of visual as well as quantitative performance measures viz. Average difference, Normalized absolute error and Pixel classification based measure. The proposed method is found to be better than other methods.