Person matching is an important topic in video- surveillance and can be used to design detection, tracking and recognition systems. Multi-scale covariance (MSCOV) is considered as one of the most promising descriptors...
详细信息
Person matching is an important topic in video- surveillance and can be used to design detection, tracking and recognition systems. Multi-scale covariance (MSCOV) is considered as one of the most promising descriptors for person matching. Unfortunately, implementing such descriptors for person matching requires heavy computation. For a system that requires real-Time matching, visual information needs to be fast processed without reducing accuracy. parallel processing is frequently adopted to speed-up execution-Time. This paper presents an optimized parallel model of a person matching system based on MSCOV. To this aim, a high-level parallelization approach based on the exploration of task and data levels of parallelism is adopted. Starting from the block diagram, an initial model that extracts the maximum task-level parallelism is proposed. This model is implemented and validated at a high system-level. Analyzing the communication and computation workload results, the potential bottlenecks of this model are identified. An optimized parallel model with the best workload balance is then developed. This model is prototyped and validated using a multicore architecture. The experimental results are promising and the system is shown to perform person matching in real-Time with 16.33 fps using a dual-core ARM-Cortex-A9.
We propose the lattice design that allows multiple topologically protected edge modes. The scattering between these modes, which is linear, energy preserving, and robust against local disorders, is discussed in terms ...
We investigate modal localization of light in disordered hyperbolic lattices. We examine modal area at the bulk of a disordered hyperbolic lattice, which demonstrates that high degree in the lattice leads to the deloc...
A two-layer continuous-time cellular neural network for finding the Radon transform of a binary image is presented. The functionality of this cellular neural network follows from the functionality of the connected com...
详细信息
A two-layer continuous-time cellular neural network for finding the Radon transform of a binary image is presented. The functionality of this cellular neural network follows from the functionality of the connected component detector cellular neural network.
We propose a perturbative design method for engineering quasi-isospectrality in multidimensional photonic systems. Our study provides platform-transparency alleviating mathematical strictness of supersymmetric transfo...
详细信息
We investigate the phase transition in developing amorphous hafnium oxides for optoelectronic applications by employing the molecular dynamic simulation. Our study provides a microscopic picture on the macroscopic opt...
We propose data-driven engineering of active light-disorder interactions. Neural networks generate the family of disorders for active multilayer structures having similar modulation sensitivity, enabling the independe...
We introduce the data-driven design of multilayers with enhanced tunability. By applying the machine-learning-designed claddings to a phase-changeable core, we obtain the deterministic realization of on-off states in ...
详细信息
Active disturbance rejection control (ADRC), as proposed by Prof. Jingqing Han, reduces first the plant dynamics to its canonical form, normally in the form of cascade integrators, for which the standard controller ...
详细信息
Active disturbance rejection control (ADRC), as proposed by Prof. Jingqing Han, reduces first the plant dynamics to its canonical form, normally in the form of cascade integrators, for which the standard controller can be employed to meet the design specifications. This paper concerns with the selection of the canonical form for non-minimum phase systems. In particular, it is shown that, by employing the well known controllable canonical form, the uncertainties of such systems can be divided into two terms in the state space model, one in the control channel and the other in the output channel. The necessary and sufficient condition is obtained for the stability of the closed-loop system with the proposed canonical form and ADRC. Also, by showing the necessity of the detectability of the extended system as well as certain information of the system-s "zeros", we present the fundamental guidelines of design ADRC for non-minimum phase uncertain systems.
暂无评论