Shape Similarity Measurement for Known-Object Localization: A New Normalized Assessment

Area of Science:

• Computer Vision
• Image Analysis
• Pattern Recognition

Background:

• Object pose estimation is crucial for robotics and computer vision.
• Existing metrics often struggle with scale, rotation, and translation variations.
• Accurate contour-based object recognition requires robust pose assessment.

Purpose of the Study:

• To introduce a new, normalized measure for contour-based object pose assessment.
• To demonstrate the robustness of the proposed measure against common shape variations.
• To provide a reliable method for known-object recognition and localization in binary images.

Main Methods:

• Development of a novel normalized performance measure.
• Quantification of differences between reference edge maps and candidate images.
• Comparative analysis against six existing object pose estimation approaches.

Main Results:

• The proposed measure shows increased robustness to translation, rotation, and scaling.
• Experimental validation on real images confirms suitability for object pose and shape matching.
• The method effectively assesses object recognition and localization accuracy.

Conclusions:

• The new normalized measure offers a more reliable approach to object pose estimation.
• This method enhances the accuracy of known-object recognition and localization.
• The proposed technique is well-suited for shape-matching applications in computer vision.