中国科学院机构知识库网格系统: 检索

CNQ: Compressor-Based Non-uniform Quantization of Deep Neural NetworksInspec keywordsOther keywordsKey words 期刊论文 OAI收割
CHINESE JOURNAL OF ELECTRONICS, 2020, 卷号: 29, 期号: 6, 页码: 1126-1133
作者: Yuan, Yong; Chen, Chen; Hu, Xiyuan; Peng, Silong

|

收藏 | 浏览/下载：90/0 | 提交时间：2021/03/08

entropy image classification learning (artificial intelligence) neural nets object detection optimisation quantisation (signal) network structure DNN low-bit quantization time-consuming training compressor-based fast nonuniform quantization method quantization model post-training quantization methods deep neural networks network quantization compressor-based nonuniform quantization CNQ Non-uniform quantization Knowledge distillation Unlabeled samples Network compression

The Cat's Eye Effect Target Recognition Method Based on Visual Attention 期刊论文 OAI收割
Chinese Journal of Electronics, 2019, 卷号: 28, 期号: 5, 页码: 1080-1086
作者: X.B.Wang; J.Zhang; S.H.Wang

|

收藏 | 浏览/下载：48/0 | 提交时间：2020/08/24

feature extraction,image fusion,object detection,target-saliency map,target region,Cat's eye effect target recognition method,visual,attention,second-directional derivative filter,directional channels,morphological method,potential targets exist,designed saliency fusing,method,candidate targets,Cat's eye effect target recognition method,based on visual attention (CTRVA),Target recognition,Saliency map,Engineering

The Cat's Eye Effect Target Recognition Method Based on Visual Attention 期刊论文 OAI收割
Chinese Journal of Electronics, 2019, 卷号: 28, 期号: 5, 页码: 1080-1086
作者: X.B.Wang; J.Zhang; S.H.Wang

|

收藏 | 浏览/下载：35/0 | 提交时间：2020/08/24

feature extraction,image fusion,object detection,target-saliency map,target region,Cat's eye effect target recognition method,visual,attention,second-directional derivative filter,directional channels,morphological method,potential targets exist,designed saliency fusing,method,candidate targets,Cat's eye effect target recognition method,based on visual attention (CTRVA),Target recognition,Saliency map,Engineering

A new approach for land cover classification and change analysis: Integrating backdating and an object-based method 期刊论文 OAI收割
REMOTE SENSING OF ENVIRONMENT, 2016, 卷号: 177, 期号: 0, 页码: 37-47
作者: Yu, Wenjuan; Zhou, Weiqi; Qian, Yuguo; Yan, Jingli

收藏 | 浏览/下载：41/0 | 提交时间：2017/03/24

Land cover classification Change analysis Backdating Object-based method Beijing

A simple and fast moving object segmentation based on H.264 compressed domain information (EI CONFERENCE) 会议论文 OAI收割
4th International Conference on Computational and Information Sciences, ICCIS 2012, August 17, 2012 - August 19, 2012, Chongqing, China
作者: Chen X.; Chen X.; Chen X.; Sun L.

收藏 | 浏览/下载：23/0 | 提交时间：2013/03/25

The paper presents a simple and fast approach for moving object segmentation based on H.264 compressed domain information for the application of indoor video surveillance with static camera. Due to the characteristics of indoor video surveillance the proposed method of segmentation avoids complicated background model like Gaussian Mixture background model. On the contrary it chooses some simple information like the type of Macroblock etc.. Experimental results of several specific H.264 compressed video sequences demonstrate the good segmentation quality of the proposed approach. 2012 IEEE.

Eight-neighborhood based background modeling (EI CONFERENCE) 会议论文 OAI收割
2012 2nd International Conference on Materials Science and Information Technology, MSIT 2012, August 24, 2012 - August 26, 2012, Xi'an, Shaan, China
Yan L.; Ming D.; Lei J.

收藏 | 浏览/下载：14/0 | 提交时间：2013/03/25

In order to solve the Gaussian kernel density-based background modeling we propose a background modeling method based on an 8-neighborhood pixels sample set. In this method we use the target pixel and its surrounding 8-neighborhood pixels to analyze whether it is included in the background or in the foreground sample space. Experimental results show that the method can judge the object as the background interference caused by the cyclical movement. By practical verification the algorithm of the background modeling can full meet the requirements of the target detection algorithm. (2012) Trans Tech Publications Switzerland.

Adaptive deformation estimation of moving target by weight image analysis (EI CONFERENCE) 会议论文 OAI收割
2010 2nd International Conference on Future Computer and Communication, ICFCC 2010, May 21, 2010 - May 24, 2010, Wuhan, China
Bai X.-G.; Dai M.

收藏 | 浏览/下载：41/0 | 提交时间：2013/03/25

An algorithm based on weight image analysis is proposed for adaptive deformation estimation of moving target in mean-shift tracking method. At the first we get the weight image from the target candidate region. Then we analyze the differences between the object and background. According to that the area estimation of the target can be converted into the image segmentation task. To realize the adaptive segmentation and estimation we define the threshold as the maximum variance between object and background. Combining the estimated area and covariance matrix we can estimate the width height and orientation of the object. The experimental results on three representative video sequences validate its robustness to the deformable estimation of the targets. 2010 IEEE.

The design of acquisition circuit for grating digital signal based on FPGA (EI CONFERENCE) 会议论文 OAI收割
2010 3rd International Conference on Advanced Computer Theory and Engineering, ICACTE 2010, August 20, 2010 - August 22, 2010, Chengdu, China
作者: Wang W.-G.

收藏 | 浏览/下载：48/0 | 提交时间：2013/03/25

In order to resolve the poor suppression capability of noise and fitter interference existing in grating encoder high-rate subdivision and the poor accuracy of kam-to counting circuit the results show that the design method will help improve the controlled object of measurement precision and control accuracy. 2010 IEEE. we design a circuit based on FPGA to realize multiplier kam and filter for the output of two-way orthogonal signal generated by Incremental Optical Encoder. The system is mainly divided into three modules such as filtering multiplier kam-to and counting. The main function of filter circuit is to eliminate the jitter and noise interference existing in the quadrate encoder signals. Kam-to multiplier circuit can accurately judge the full cycle and half-cycle of incremental encoder at the same time can make fourfold multiplier. Counting circuit can use IP cores owned by Quartus II which is not restricted on the median. At last timing simulation based on Modelsim carried on the three modules

A real-time two-dimensional correlation speed measurement based on image (EI CONFERENCE) 会议论文 OAI收割
2010 International Conference on Computer, Mechatronics, Control and Electronic Engineering, CMCE 2010, August 24, 2010 - August 26, 2010, Changchun, China
作者: Yu L.

收藏 | 浏览/下载：36/0 | 提交时间：2013/03/25

In order to implement two-dimensional moving object's non-contact and real-time speed measurement in this paper wc proposed a new method which was based on the characteristic of FPGA and a real-time two-dimensional correlation speed measurement was proposed on image processing using high speed CMOS shown in figure 1. Experimental installation was shown in figure 2 a cardboard with images which was fixed in electric control movement device was moving to simulate moving object the CMOS image sensor gathered image signals of the cardboard and the image signals was output to FPGA after a serials pretreatment such as binaryzation etc. Logical operation insteaded mathematical operation on FPGA to implement two frames image's cross-correlation operations as the formula (10) in this paper and finally calculate the two-dimensional velocity. Through experiment result of table 1 we drawn an conclusion the accurarcy of the method in this paper was less than 1 pixel and the method can be used at a certain speed ranges of two-dimensional real-time speed measurement. 2010 IEEE.

Study particle filter tracking and detection algorithms based on DSP signal processors (EI CONFERENCE) 会议论文 OAI收割
2010 International Conference on Computer, Mechatronics, Control and Electronic Engineering, CMCE 2010, August 24, 2010 - August 26, 2010, Changchun, China
Dong Y.; Chuan W.

收藏 | 浏览/下载：32/0 | 提交时间：2013/03/25

In Video tracking detection and tracking usually need two algorithms. The process is complex and need much time which detection and tracking are. In this paper a hybrid valued sequential state vector is formulated. The state vector is characterized by information of target appearance flag and of location. Particle filter-based method implements detection and tracking at one time. In order to reduce process time and think of pixel position in tracking field feature histogram of luminance is as observe vector and used posterior estimate. In this paper the luminance component is derived and target is recognized and tracked through image processor based on DSP in order to implementing real-time. The experimental results confirm that method can detect and track the object in real-time successfully when the number of particles is 160. The method is robust for rolling scale and partial occlusion. 2010 IEEE.