Transformation of image positions, rotations, and sizes into shift parameters

Area of Science:

• Optics and Photonics
• Image Processing
• Robotics

Background:

• Classical optical correlators struggle with variations in object scale and rotation.
• Simultaneous analysis of multiple objects requires efficient feature extraction.

Purpose of the Study:

• To develop an optical image processing system that encodes orientation and size into shift properties.
• To enable rotation and size-invariant object recognition using multiplexed filters.
• To create an optical robot vision system capable of multi-object analysis.

Main Methods:

• Analytical and experimental description of the optical image processing system.
• Utilizing a classical correlator with a rotation and size-invariant multiplexed match filter.
• Implementing a two-measurement approach at different rotation angles for scene analysis.

Main Results:

• The system successfully converts orientation and size to shift properties while preserving positional information.
• The transformed image is compatible with classical correlators for further processing.
• Demonstrated potential for simultaneous multi-object analysis in robot vision.

Conclusions:

• The proposed optical system offers a novel method for encoding geometric object properties into shift parameters.
• This approach facilitates robust and efficient object recognition in robotic applications.
• The system's ability to analyze multiple objects simultaneously with minimal measurements is a significant advancement.