Occluded objects recognition using multiscale features and Hopfield neural networks

Jiann Shu Lee, Chin Hsing Chen, Yung Nien Sun, Guan Shu Tseng

Research output: Chapter in Book/Report/Conference proceedingConference contribution


A new method to recognize partially visible two-dimensional objects by means of multiscale features and Hopfield neural network is proposed. The Hopfield network is employed to perform global feature matching. Since the network only guarantee to converge to a local optimal state, the matching results heavily depend on the initial network state determined by the extracted features. To acquire more satisfactory initial matching results, a new feature vector, consisting of the multiscale evolution of the extremal position and magnitude of the wavelet transformed contour orientation, is developed. These features contain the contour primitives information in a multiscale manner, hence good initial states can be obtained. The good initiation enables the method to recognize objects of even heavily occluded, that can not be achieved by using the Nasrabadi-Li’s method. In addition, to make the matching results more insensitive to the threshold value selection of the network, we replace the step-like thresholding function by a ramp-like one. Experimental results have shown that our method is effective even for noisy occluded objects.

Original languageEnglish
Title of host publicationImage Analysis and Processing - 8th International Conference, ICIAP 1995, Proceedings
EditorsCarlo Braccini, Leila DeFloriani, Gianni Vernazza
PublisherSpringer Verlag
Number of pages6
ISBN (Print)3540602984, 9783540602989
Publication statusPublished - 1995
Event8th International Conference on Image Analysis and Processing, ICIAP 1995 - San Remo, Italy
Duration: 1995 Sep 131995 Sep 15

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Other8th International Conference on Image Analysis and Processing, ICIAP 1995
CitySan Remo

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)


Dive into the research topics of 'Occluded objects recognition using multiscale features and Hopfield neural networks'. Together they form a unique fingerprint.

Cite this