Video target tracking is an important research topic in computer vision, and has been widely used in video surveillance, robot, human-computer interaction and so on. The emergence of large data age and the emergence o...
详细信息
Video target tracking is an important research topic in computer vision, and has been widely used in video surveillance, robot, human-computer interaction and so on. The emergence of large data age and the emergence of in-depth learning methods provide a new opportunity for the study of video target tracking. This paper first analyzes the research problems of video target tracking at present, analyzes the characteristics and trends of video target tracking in the new period, introduces the emerging recursive neural network frame structure, combined with Kalman filter And the experimental results show that the accuracy and robustness of the target tracking based on the convolution neural network algorithm are all good.
In the process of image acquisition and transmission, the image always generates noise due to internal and external interference. Noise reduces the quality of the image, and makes it difficult for subsequent image pro...
详细信息
In the process of image acquisition and transmission, the image always generates noise due to internal and external interference. Noise reduces the quality of the image, and makes it difficult for subsequent image processing. Therefore, image denoising is very important in image processing. Wavelet denoising can effectively filter out noise and retain high-frequency information of the image, this method has the characteristics of fast operation speed and has become an important branch of image denoising. Threshold functions commonly used in wavelet threshold denoising include hard threshold function and soft threshold function. The hard threshold function is not continuous as a whole. Although the soft threshold function has good continuity, there is always a constant deviation between the processed coefficient and the original coefficient when the wavelet coefficient is large. In response to these deficiencies, this paper establishes a new improved threshold function based on traditional soft and hard threshold functions. By processing the thresholds of wavelet coefficients, a reasonable balance between smoothing and edge oscillations can be achieved after image denoising. The improved threshold function not only overcomes the shortcomings of the soft and hard threshold functions, but also provides more flexibility in the processing of image *** MATLAB simulation, the denoising effects of the soft, hard threshold functions and the threshold function constructed in this paper are compared in terms of signal-to-noise ratio(SNR) and root mean square error(MSE). The MATLAB simulation results show that compared with the traditional threshold function, the improved threshold function has a higher signal-to-noise ratio(SNR = 26.27709) and a smaller mean square error(MSE = 153.4579), and it has a good noise reduction effect.
A piecewise control strategy is proposed to realize the position-posture control of a planar four-link Active-PassiveActive-Active(APAA) underactuated manipulator(UM). Specially, the particle swarm optimization(...
详细信息
A piecewise control strategy is proposed to realize the position-posture control of a planar four-link Active-PassiveActive-Active(APAA) underactuated manipulator(UM). Specially, the particle swarm optimization(PSO) algorithm is used to obtain the target angle of all links based on the constraints of control objects of the system. The overall control process of the system is divided into two stages: Firstly, we design a fuzzy-PI controller for the first link to realize the control target of the passive link, and the error between the current angle and the desired angle of the passive link is converged to zero by adjusting the velocity of the first link. Secondly, the position mode of the servo controller is adopted to control the active links move to their target angles, respectively. Finally, the experimental results of the real system verify the effectiveness of the proposed control strategy.
To ensure the efficiency of people’s work while working long hours,an initiative service method with regard to degree of sleepiness for drinking service robot is *** can recognize the degree of sleepiness and provide...
详细信息
To ensure the efficiency of people’s work while working long hours,an initiative service method with regard to degree of sleepiness for drinking service robot is *** can recognize the degree of sleepiness and provide initiative service to increase the efficiency of people’s *** of sleepiness is defined as three levels by observer rating,which is supposed to be associated with human actions,environmental factors,and individual *** addition,a model of degree of sleepiness based on Finite State Machines(FSM) is *** relationship between human demands and degree of sleepiness is established through demand analysis model by which the initiative service with regard to degree of sleepiness for drinking service robot is *** on real initiative drinking service are performed in a multimodal emotional communication based on human-robot interaction(MEC-HRI) system,from which the experimental results show that the time of initiative service is 8 minutes earlier than that without initiative *** prospect,the proposal could be applied to other kinds of services such as health care,safety driving assistance,and other safety work assistance.
An important feature of the deep learning algorithm is that the hidden layer of the neural network is more dependent on more computing resources and a larger amount of data. In this paper, we use Triplet GAN method to...
详细信息
An important feature of the deep learning algorithm is that the hidden layer of the neural network is more dependent on more computing resources and a larger amount of data. In this paper, we use Triplet GAN method to identify human body images with tattoos using fewer data sample labels. The model achieved better results(0.9462) than the only Triplet(0.8352) and only GAN(0.9178) in the MNIST dataset, it also achieved higher recognition accuracy(77-82%) under the premise of saving the amount of data, compared with the traditional machine learning algorithm(54-84%). We provide a new idea for tattoo image recognition applications in embedded computing units.
In this paper,a simple one-stage control strategy is developed based on optimization algorithm to quickly realize the position control of a planar n-link(n ≥ 3) underactuated manipulator with a passive first *** co...
详细信息
In this paper,a simple one-stage control strategy is developed based on optimization algorithm to quickly realize the position control of a planar n-link(n ≥ 3) underactuated manipulator with a passive first *** continuous control of a planar three-link underactuated manipulator with a passive first joint can be realized,and so can the planar n-link(n ≥ 3) ***,one problem that the number of the variables needing to be optimized will increase with the increase of the number of the links will ***,considering the planar three-link manipulator can be controlled to reach to any target position,we employ the differential evolution(DE) algorithm to optimize the angles of only two active links and the design parameters of the corresponding ***,the angles and the corresponding design parameters of other n — 3 active links are kept being *** this way,all link angles can simultaneously converge to their target *** a planar four-link underactuated manipulator with a passive first joint as an example,the simulation results verify the effectiveness and rapidity of the proposed strategy.
In order to avoid the linear inversion method falling into local minima and slow convergence speed of the global optimization inversion method, the article proposed the simplex-simulated annealing algorithm for transi...
详细信息
In order to avoid the linear inversion method falling into local minima and slow convergence speed of the global optimization inversion method, the article proposed the simplex-simulated annealing algorithm for transient electromagnetic inversion research which combines advantages of the simplex method and the simulated annealing algorithm. The simplex method is used to obtain local minimum value which is relatively close to the actual value, then the simulated annealing algorithm is used to obtain the global optimal solution which can better reflect structural characteristics of the real stratigraphic *** the comparison of the noise inversion results and noise free inversion results about K-type, H-type, KH-type and HKH-type stratigraphic models, it can be proved that the simplex-simulated annealing algorithm can suppress some noise. The comparison of the simulated annealing method and the simplex-simulated annealing algorithm shows that the simplex-simulated annealing algorithm has the characteristics of global searching ability and fast convergence speed.
This paper suggests a novel technique for the tool parameter measurement based on machine vision. Tool images are captured by using a machine vision system and the outer contour image of the cutter is obtained by usin...
详细信息
This paper suggests a novel technique for the tool parameter measurement based on machine vision. Tool images are captured by using a machine vision system and the outer contour image of the cutter is obtained by using the machine vision technology. The HALCON image processing library is used as the development platform to build the tool parameters test system. Several algorithms including image segmentation, edge extraction and fitting ellipse determination is used for image processing. The tool parameters such as external diameter and contour angle of tool edge can be obtained after rebuilding the contour of tool edge. The proposed scheme is shown to be reliable and effective for the automated tool parameter measurement.
Facial expression recognition(FER) plays an important role in human-machine interaction. An assistant robot having a close interaction with human being should be able to recognize human facial expression. FER is a non...
详细信息
Facial expression recognition(FER) plays an important role in human-machine interaction. An assistant robot having a close interaction with human being should be able to recognize human facial expression. FER is a non-trivial problem because each individual has his own way to reveal his emotion and the facial expressions of two different persons may not be totally identical. Hence,facial expression recognition is still a challenging problem in computer vision. In this work, we propose a simple solution for facial expression recognition that uses a combination of Convolutional Neural Network and specific image pre-processing *** experiments employed to evaluate our technique were carried out using two largely used public databases(CK+, JAFFE).A study of the impact of each image pre-processing operation in the accuracy rate is presented. The proposed method: achieves competitive results when compared with other facial expression recognition methods-97.85% of accuracy in the CK+ database-it is fast to train,and it allows for real time facial expression recognition with standard computers.
Musical staff lines detection and removal, the first step of most Optical Music Recognition(OMR) technology, aims to detect staff positions and segment score image by removing those staff lines. To deal with issues ...
详细信息
Musical staff lines detection and removal, the first step of most Optical Music Recognition(OMR) technology, aims to detect staff positions and segment score image by removing those staff lines. To deal with issues that the disappearance of thin staff lines in the captured score image, line repaired algorithm based on music rules is proposed in this paper. It first estimates two reference lengths, staff width and space between two adjacent lines. Then projection is used to obtain potential staves *** least the staves are finally determined based on line repaired algorithm, the missing lines repaired and the redundant lines deleted at the same time. The proposed technique is simple and effective, making full use of global information of the musical scores. Experimental results show that our method achieves impressive results on music score images captured from cameras.
暂无评论