An incremental learning algorithm using RBF networks is proposed. When learning new patterns, old knowledge is reserved by retrieving interfered patterns. Simulated experiments proves the correctness and effectiveness...
详细信息
An incremental learning algorithm using RBF networks is proposed. When learning new patterns, old knowledge is reserved by retrieving interfered patterns. Simulated experiments proves the correctness and effectiveness of the algorithm.
The 3-D texture properties of the inner surface of the engine cylinder were converted and the method for wavelet analysis could be applied to extract the parameters of the surface. The characteristics of inner surface...
详细信息
The 3-D texture properties of the inner surface of the engine cylinder were converted and the method for wavelet analysis could be applied to extract the parameters of the surface. The characteristics of inner surface were decomposed and reconstructed by the module maximum of wavelet transfer. The desired topography parameters were obtained. The experimental results showed that the method could be used to measure the honing parameters accurately and was helpful to improve the technology.
The real time images were fuzzy because of the hypsography in radar scene matching. The positioning results of matching did not coincide with the true results. A method was proposed to adjust the errors caused by the ...
详细信息
The real time images were fuzzy because of the hypsography in radar scene matching. The positioning results of matching did not coincide with the true results. A method was proposed to adjust the errors caused by the positioning results of matching by the influence by means of analyzing the influence of the hypsography on the real time images. It was proved that the method was effective.
The application of rough sets combined with neural network method in target recognition by data fusion is developed in this paper. The learning mechanism has been introduced into the rough sets system and the neural n...
详细信息
The application of rough sets combined with neural network method in target recognition by data fusion is developed in this paper. The learning mechanism has been introduced into the rough sets system and the neural network is built up according to the attribute of conditions and the decision rules of the rough sets. Finally, the experiments with three different kinds of target images in three different kinds of spectrums shows that the combination of rough sets and neural network get a much higher recognition rate than taking one data fusion algorithm alone, also, the training time is reduced in large scale.
Based on the crosstalk faults of interconnects between IP cores in SOC, this paper presented an implementation method for the model by using the control and operation function of processor core in the SOC. The method ...
详细信息
Based on the crosstalk faults of interconnects between IP cores in SOC, this paper presented an implementation method for the model by using the control and operation function of processor core in the SOC. The method for software implementation needed less hardware burden in comparison with the traditional implementation one based on the hardware. A new IP core framework to improve the detecting speed of crosstalk fault was presented. The improved IP core made the test finished in the real-time and parallel requirement.
A new path planning method of mobile robot in unknown environment using self-learn visual graph is proposed. In this paper, a presentation named by self-learn visual graph (SLVG) is used to represent unknown environme...
详细信息
A new path planning method of mobile robot in unknown environment using self-learn visual graph is proposed. In this paper, a presentation named by self-learn visual graph (SLVG) is used to represent unknown environment, the SLVG is empty at first, when the path planning is conduct, the SLVG is built partly, when the path planning is finished, the SLVG is built completely. On the base of SLVG, a local optimal obstacle avoidance algorithm and a local optimal path planning algorithm are also present in the paper. Experimental results using simulation demonstrate that the method is fast and can produce a local optimal path for mobile robot in unknown environment.
A method was proposed to improve the target recognition quality by combining the physical constrains or prior knowledge as evidences into target recognition within the frame of mathematical statistic theory and Dempst...
详细信息
A method was proposed to improve the target recognition quality by combining the physical constrains or prior knowledge as evidences into target recognition within the frame of mathematical statistic theory and Dempster-Shafer's evidence theory. In this method, the usability of the evidences was appraised with Kolmogorov-Smirnov test method and the different computation models to compute the belief value to classifier's result corresponding to the different evidence types proposed. The method was tested on the real infrared images sequences with complex background, reducing the false alarm from 0.031 to 0.008 and miss alarm from 0.085 to 0.038 respective, that is, 3/4 and 1/2 times respectively. The result indicates that the proposed method for improving the recognition performance is feasible and effective.
This paper proposes a genetic-based algorithm for surface reconstruction of three-dimension (3-D) objects from a group of contours representing its section plane lines. The algorithm can optimize the triangulation of ...
详细信息
This paper proposes a genetic-based algorithm for surface reconstruction of three-dimension (3-D) objects from a group of contours representing its section plane lines. The algorithm can optimize the triangulation of the surface of 3-D objects with a multi-objective optimization function to meet the needs of a wide range of applications. Further, a new crossover operator for triangulation and a new 3-D quadrilateral mutation operator are also introduced.
A regularized restoration algorithm based on maximum-likelihood estimation was presented for restoring object images from the noisy turbulence-degraded images. The logarithmic maximum-likelihood function for multi-fra...
详细信息
A regularized restoration algorithm based on maximum-likelihood estimation was presented for restoring object images from the noisy turbulence-degraded images. The logarithmic maximum-likelihood function for multi-frame image data based on the model of image random field was built, and some auxiliary terms to smooth noise while preserve the edges of images and the penalized item to avoid trivial solutions were added to the maximum-likelihood function. The iterative formulas of calculating the PSFs and object image were derived so that the PSFs and the object image could be estimated in the iterative manner. A parallel processing scheme for the algorithm is also proposed. The restoration experiments on the simulated turbulence-degraded images in the case of noise show that the proposed algorithm has high ability of noise-resisting and it has some practical applications.
For the image processing system which consists of multiple DSPs-ADSP21060, an embedded system software based on the micro-kernel provided by Virtuso 4.2 is designed. The implementation of the task schedule, resources ...
详细信息
For the image processing system which consists of multiple DSPs-ADSP21060, an embedded system software based on the micro-kernel provided by Virtuso 4.2 is designed. The implementation of the task schedule, resources management, parallel processing, interrupt acknowledgment and data communication in the system is described in this paper. In terms of some image processing algorithms, the performance of the system is tested. It showed that the system software proposed is easy to realize and develop. It is also steady and reliable and therefore this approach is suggested to be used widely in the image processing system.
暂无评论