Object Matching by Image Contours Using Neural Networks. | ||||
| MEJ- Mansoura Engineering Journal | ||||
| Article 7, Volume 19, Issue 4, December 1994, Page 75-91 PDF (1.21 MB) | ||||
| Document Type: Research Studies | ||||
| DOI: 10.21608/bfemu.2021.164258 | ||||
 View on SCiNiTO | ||||
| Authors | ||||
| A. El-Shami* 1; H. Niemann2; A. Tolba3 | ||||
| 1Department of Mathematics., University of Suez Canal., Egypt. | ||||
| 2Department of Computer Science., University of Erlangen., Germany. | ||||
| 3Department of Electrical Engineering., University of Suez Canal., Egypt. | ||||
| Abstract | ||||
| This paper deals with the implementation of 3-d object matching system. The shape of the object is identified from the image lines and curves using Hough transform, chain code and backpropagation neural networks. This is achieved by first dynamically thresholding the grey level image, then segmenting the image into its linear components with both Hough transform and chain coding. A backpropagation framework is used for classifying the image into one of possible surfaces based on the extracted vertices and line segments. To fix the number of input layer neurons, the image features are normalized. The approach is tried on a variety of real objects and appears to hold great promise. | ||||
| Keywords | ||||
| 3-D Object recognition; Shape matching; Chain code; Hough transform; Surface classification; Neural networks | ||||
|
Statistics Article View: 67 PDF Download: 129 |
||||
