Continual progress in the fields of computervision and machine learning has provided opportunities to develop automatic tools for tagging images;this facilitates searching and retrieving. However, due to the complexi...
详细信息
We propose a framework for efficient storing and scalable browsing of surveillance video based on video synopsis. Our framework employs a novel synopsis analysis scheme named Detail-based video synopsis to generate a ...
详细信息
We propose a framework for efficient storing and scalable browsing of surveillance video based on video synopsis. Our framework employs a novel synopsis analysis scheme named Detail-based video synopsis to generate a set of object flags to store and browse surveillance video synopsis. The main contributions of our work are: 1) highlighting important contents of surveillance video; 2) improving the storage efficiency of original video and synopsis video; 3) realizing multi-scale scalable browsing of synopsis video while reserving essential information. The experiments of implementing the framework are shown compared with the previous independent storage method of original video and synopsis video.
We present a physics-based approach for illuminant color estimation of arbitrary images, which is explicitly designed for handling images with multiple illuminants. The majority of techniques that extract the illumina...
详细信息
We present a physics-based approach for illuminant color estimation of arbitrary images, which is explicitly designed for handling images with multiple illuminants. The majority of techniques that extract the illuminant color assume that the illumination is constant across the scene. This, however, is not often the case. We propose an illuminant-color estimation method which is based on robust local illuminant estimates. There are no assumptions on the number or type of illuminants. An illuminant color estimate is obtained independently from distinct image mini-regions. From these mini-regions a robust local illumination color is computed by consensus. These local estimates are then used in deriving the chromaticity of the dominant illuminants. Experiments on an established benchmark database of real-world images show that our technique performs comparably to uniform-illuminant estimation methods. Furthermore, extensive tests on real-world images show that we can reliably process mixed illuminant scenes.
The contour analysis and identification are the important aspects in visual surveillance research. The paper proposes a fuzzy identification method of contours. First, according to the description of a contour based o...
详细信息
Pedestrian classification is addressed by T. Watanabe et al. using SVM with 34704 CoHOG features. This paper addresses the pedestrian classification using neural network with 1344 CoHOG features (feature size is 25 ti...
详细信息
Pedestrian classification is addressed by T. Watanabe et al. using SVM with 34704 CoHOG features. This paper addresses the pedestrian classification using neural network with 1344 CoHOG features (feature size is 25 times small) and still achieve comparable results.
The current paper proposes a new parametric local color correction technique. Initially, several color transfer functions are computed from the output of the mean shift color segmentation algorithm. Secondly, color in...
详细信息
The current paper proposes a new parametric local color correction technique. Initially, several color transfer functions are computed from the output of the mean shift color segmentation algorithm. Secondly, color influence maps are calculated. Finally, the contribution of every color transfer function is merged using the weights from the color influence maps. The proposed approach is compared with both global and local color correction approaches. Results show that our method outperforms the technique ranked first in a recent performance evaluation on this topic. Moreover, the proposed approach is computed in about one tenth of the time.
This paper proposes a novel method based on bag of local space-time interest points' features to recognize and retrieval complex events in real movies. In this method, an individual video sequence is represented a...
详细信息
This paper proposes a novel method based on bag of local space-time interest points' features to recognize and retrieval complex events in real movies. In this method, an individual video sequence is represented as a bag of local space-time features then we integrate such bag-of-feature with SVM for recognition events. Local space-time features are introduced to capture the local events in video and can be adapted to size and velocity of the pattern of the event. To evaluate effectiveness of this method, this paper uses the public Hollywood dataset, in this dataset the shot sequences has collected from 32 different Hollywood movies and it includes 8 event classes. The presented result justify the proposed method explicitly improve the average accuracy and average precision compared to other relative approaches.
Three-dimensional metrology problems are traditionally the domain of expensive devices like Laser-stripe sensors and custom-made projector-camera systems. The launch of Microsoft's Kinect sensor solved many of the...
详细信息
Three-dimensional metrology problems are traditionally the domain of expensive devices like Laser-stripe sensors and custom-made projector-camera systems. The launch of Microsoft's Kinect sensor solved many of these problems which were deemed feasible only with several thousand-Dollar devices before. However, the sensor is highly tuned to its domain of application, which is the robust scanning of indoor environments at an accuracy range of few centimeters. In this work we investigate how the sensor can be modified to work at much higher accuracy, on a limited but scalable measurement range. By altering the sensor baseline and depth of field, we are able to reliably retrieve depth fields of objects at an accuracy in the sub-millimeter range. We further improve the sensor by adding multiple projectors, thus creating a low-cost multi-projector, single-camera system working at a frame rate of 30 Hz, at virtually no CPU consumption. We are hereby able to apply automated pattern intensity adaption and multi-view depth fusion in real-time and compete with a large range of sensors in terms of robustness, speed and accuracy, but at a fraction of the costs.
Reconfigurable hardware such as FPGAs are being increasingly employed for application acceleration due to their high degree of parallelism, flexibility and power efficiency - factors which are key in the rapidly evolv...
详细信息
Reconfigurable hardware such as FPGAs are being increasingly employed for application acceleration due to their high degree of parallelism, flexibility and power efficiency - factors which are key in the rapidly evolving field of embedded real-time vision. While recent advances in technology have increased the capacity of FPGAs, lack of standard models for developing custom accelerators creates issues with scalability and compatibility. In this paper, we describe a model for designing streaming hardware accelerators with run-time configurability. This model provides a generic interface for each hardware module, a modular and hierarchical structure for parallelism at multiple levels and a run-time reconfiguration framework for increased flexibility. We present case studies to accelerate sample neu-romorphic vision algorithms which are inspired by models of the mammalian visual cortex. Experimental results show speedups of several factors over comparable CPU implementations and higher performance-per-watt over relevant GPU implementations.
Face detection in low resolution images is a huge challenge in computervision due to its insufficient information. In this paper, an improved LBP for low resolution facial image feature extraction is proposed which c...
详细信息
Face detection in low resolution images is a huge challenge in computervision due to its insufficient information. In this paper, an improved LBP for low resolution facial image feature extraction is proposed which considers both partial and holistic facial features. Extensive experiments show that the proposed method has an encouraging performance with good robustness and strong practicality in face detection.
暂无评论