Learning Structural Descriptions From Examples
Learning Structural Descriptions From Examples
Computer perception of complex patterns
IJCAI'71 Proceedings of the 2nd international joint conference on Artificial intelligence
Building a Distance Function for Gestalt Grouping
IEEE Transactions on Computers
A Versatile Machine Vision System for Complex Industrial Parts
IEEE Transactions on Computers
Locating Object Boundaries in Textured Environments
IEEE Transactions on Computers
Boundary and object detection in real world images
IJCAI'75 Proceedings of the 4th international joint conference on Artificial intelligence - Volume 1
| Hi-index | 14.99 |
The problem considered is the development of a method of scene analysis that employs a descriptive approach in the analysis of pictures. In an effort to facilitate implementation of the algorithm to a wide range of picture classes, a description of the scene is read in as data in the form of a tree structure that guides the search for objects from the largest to the smallest. The algorithm has the following features: the analysis of the scene is top-down; feature extraction and pattern recognition are combined in a reinforcing system; the algorithm contains a description of the class of scenes to be analyzed; region enumeration techniques are used in primitive object identification; principal parameters in the program are self-adjusting; and the concept of ``field of vision'' is used to locate the boundaries of objects. The algorithm has been implemented and tested on posteroanterior (PA) chest radiograms, anteroposterior (AP) knee radiograms, and lateral brain scans. The analysis of PA chest radiograms is used to describe the implementation of the program. To date, the algorithm has been tested only on radiographic images; however, the method should also be applicable to other image classes.