Journal of Northwestern Polytechnical University Apr. Vol No. 2 ICM ICM ICM ICM ICM ICM TP A

Size: px

Start display at page:

Download "Journal of Northwestern Polytechnical University Apr. Vol No. 2 ICM ICM ICM ICM ICM ICM TP A"

裙英牧
5 years ago
Views:

1 Journal of Northwestern Polytechnical University Apr. Vol No. 2 ICM ICM ICM ICM ICM ICM TP A Kinser Johnson ICM intersecting cortical model 7 ICM ICM Hu 1 Fourier 2 Trace 3 ICM Hu Hu Fourier Trace ICM Radon 1 ICM 4 ICM ICM M N 1990 M N M N ICM Eckhorn N 5 ij 1993 Johnson ckhorn VLSI JQ

2 F ij n + 1 = ff ij n + S ij + M ijkl Y kl n 1 kl n T ij n + 1 = gt ij n + hy ij Y ij n + 1 = 1 F ij n + 1 T ij n { 3 0 otherwize G n = Y ij n 4 0 R c 1 i j S ij N ij Y ij 5 N ij F ij M ijkl f g h G n 2. 2 ICM n N ij ICM 1 ~ 4 ICM 7 ICM 8 M ijkl f g h M ijkl ICM M ijkl f g h ICM a f g h f 1 g 1 h 1 b E = 1 2 R c R c c < ξ f 1 g 1 h 1 ICM ICM 2 ICM f R c ICM h = f g h + Δh - E f g h 7 Δh λ 1 = E f g h λ ICM 2 2 = E f g h λ i = 1 ) 2 3 = E f g h f i = 1 ) 2 g i = 1 h ) E 1 m+1 = E 1m - λ 1 E ICM f 2 m+1 = E 2m - λ 2 g E 3 m+1 = E 3m - λ 3 9 h 1 b c E = 1 2 R 10 c λ k + 1 = R c = G s j G g j [ G 2 s j ] 1 /2 [ G 2 g j ] 1 /2 f + Δf = g h - E f g h Δf g = f g + Δg h - E f g h Δg 6

2 ICM 203 { λ k otherwise 1. 05λ k E k + 1 < E k 0. 75λ k E k + 1 > 1.

3 2 ICM 203 { λ k otherwise 1. 05λ k E k + 1 < E k 0. 75λ k E k + 1 > E k 10 ICM 3 ICM ICM ICM f g h 2 1a 1b ICM 11 N F ω - F' ω 2 ω = 1 D = 11 N F ω 2 ω = 1 F ω F' ω N ~ c 1d f = 0. 6 g = 0. 1 h = 40 1a 1b ICM 1e 1f f = 0. 4 g = 0. 1 h = 60 1a 1b ICM ICM ICM f 3 15 = 0. 9 g = 0. 8 h = 20 ICM 2

4 F-22 F-35 Saab B2 F-15 F-16 F-18 Su Mig-29 F ICM ICM ICM ICM ICM Sun Jixiang. Feature Extraction in Pattern and Computer Vision Invariants. Beijing National Defence Industry Press 2001 in Chinese

5 2 ICM Kauppien H Sepanen T. An Experiment Comparison of Autoregressive and Fourier Based Descriptors in 2D Shape Classification. IEEE Trans on Pattern Analysis and Machine Intelligence Kadyrov A Petrou M. The Trace Transform and Its Application. PAMI Wang Fenguo. Wavelet Transform and Its Application in Affine Invariant Object Recognition. Master Degree Dissertation. Xi'an University of Electronic Science and Technology 2007 in Chinese 5 ckhorn R Reitboeck H J Arndt Metal. Feature Linking Via Synchronization among Distributed Assemblies Simulation of Results from Cat Cortex. Neural Comput Johnson J L Ritter D. Observation of Periodic Waves in a Pulse-Coupled Neural Network. Opt Lett Kinser J M. Image Signatures Classification and Ontology. Proc IASTED Int Conf on Computer Graphics and Imaging Yu Ruixing Zhu Bing Zhang Ke et al. An Image Matching Recognition Approach Based on ICM Icon / FSKD Ulf Ekblad Jason M Kinser. Theoretical Foundation of the Intersecting Cortical Model and Its Use for Change Detection of Aircraft Cars and Nuclear Explosion Tests. Signal Processing Xu Xin He Hangen. A Gradient Algorithm for Neural-Network-Based Reinforcement Learning. Chinese Journal of Computers in Chinese Luo Min Zhu Xiaomin Li Xiaohong. Wavelet-Radial Descriptor Based Algorithm for Image Feature Extraction. Geomatics and Information Science of Wuhan University in Chinese A New and Effective Method for Setting Parameters of ICM Intersecting Cortical Model Yu Ruixing Meng Lixun College of Astronautics Northwestern Polytechnical University Xi'an China Abstract In identifiying targets the parameters of the ICM model need to be set manually. So we propose what we believe to be a new and effective method for setting the parameters which is explained in sections 1 and 2 of the full paper. Section 1 briefs the ICM. Subsections 2. 1 deals with the error function which is based on correlation coefficients and the ICM parameteer setting procedures based on gradient descent subsection 2. 2 gives a 3-step calculation procedure for adaptively setting the ICM parameters in accordance with different input images. Section 3 uses three numerical examples to verify the effectiveness of our method the simulation results given in Figs. 1 through 3 and Tables 1 and 2 and their analysis show preliminarily that our new method has slightly higher target recognition rate than the method mentioned in Ref. 11 and can indeed effectively set the ICM parameters and enhance the reliability and efficiency of the ICM. Key words algorithms analysis calculations correlation methods effects efficiency feature extraction functions image processing models numerical methods optimization parameter estimation reliability simulation standards targets error function gradient descent intersecting cortical model ICM target recognition icon

1 引言

1 引言 P P 第 40 卷 Vol.40 第 7 期 No.7 计算机工程 Computer Engineering 014 年 7 月 July 014 开发研究与工程应用文章编号 :1000-348(014)07-081-05 文献标识码 :A 中图分类号 :TP391.41 摘基于图像识别的震象云地震预测方法谢庭,