摘要
脉冲耦合神经网络的时间序列在图像检索和识别中应用广泛,但是时间序列无法体现图像的形状特征,造成图像判别失败。提出交叉视觉皮层的局部时间序列来解决上述问题。首先将图像分块,然后分别提取图像各部分的时间序列,最后将其连接形成整体的时间序列。提出的算法与基本的时间序列及加入边缘信息的时间序列比较,实验证明该方法解决了基本时间序列存在的问题,同时算法效率和准确率更高。
The time series of Pulse Coupling Neural Network（PCNN） is widely used in the image retrieval and identification,but it cannot embody the shape and characteristics of the image,which results in the failure of image evaluation.In this paper,the local time series of cross visual cortex was proposed to solve the problem.Fist,the image was divided into blocks;then,the time series of each block was extracted;last,the local time series were linked to global time series.The proposed algorithm was compared with the basic time series and the time series added with edges information.The experimental results demonstrate that the proposed method can effectively and efficiently solve the problems existing in the basic time series.
出处
《计算机应用》
CSCD
北大核心
2011年第6期1588-1591,共4页
journal of Computer Applications
基金
国家自然科学基金资助项目(60970098
60803024)
国家自然科学基金重大研究计划项目(90715043)
教育部高等学校博士点基金资助项目(20090162110055)
新教师基金资助项目(200805331107)
浙江大学CAD&CG国家重点实验室开放项目(A1011A0911)
湖南省教育厅科研资助项目(09C745)
关键词
图像判别
脉冲耦合神经网络
交叉视觉皮层
时间序列
交通标志
image identification
Pulse Coupled Neural Network（PCNN）
intersecting cortical model
time series
traffic sign
