Conventional Technology

Pattern matching is a simple yet powerful machine vision technology. The normalized correlation method has been widely used in machine vision applications because the match score is largely independent of linear variations in object shading caused by reflection or illumination variations. However, the normalized correlation method does not work well when the pattern being searched is subject to rotation or size variation. The match score generally drops significantly when only a few degrees of rotation or a small percentage of size change occurs.

Geometric pattern matching approaches have been developed to resolve the above limitations. They use geometric information in place of pixel grid-based normalized correlation for matching. Key geometric features are extracted within an object image using methods such as boundary encoding. Then the spatial relationships in the pattern template are adjusted (by scaling and rotation) to match the key features of the input image. However, this approach requires high edge contrast and low noise between patterns and background to reliably extract the key geometric features for matching. The geometric pattern matching approach fails when edges of the key features are determined incorrectly due to noisy or indefinite input image edges.

DRVision Technologies LLC technology is more robust and also computationally efficient

DRVision Technologies LLC invariant matching technology is a fast pattern search method that can accurately locate patterns of interest in instances where they vary in size or orientation, when their appearance is degraded, and even when they are partially hidden from view. It retains the advantage of the pixel grid based normalized correlation approach on low contrast and noisy images and it achieves the additional advantages of rotation and scale invariance.

DRVision Technologies LLC technology allows for fast linear search of scale and a fast linear search for rotation. In addition, DRVision Technologies LLC technology includes partial search and integration to handle significant image degradation or target occlusion. A coarse to fine search process increases the scale and rotation search speed. Wide search ranges are applied using lower resolution images and a fine-tuning search uses higher resolution images. This efficiently achieves both wide search range and fine search resolution.