Hysteresis thresholding

Abstract

Hysteresis thresholding offers enhanced object detection but is time consuming, requires lots of memory resources, and is unsuitable for VSNs. In this chapter, we present a unified compact architecture that couples Hysteresis Thresholding with connected component analysis and Object Feature Extraction (HT-OFE) in a single pass over the image. Two versions are developed: a high-accuracy pixel-based architecture and a faster block-based one at the expense of some accuracy loss. Unlike queue-based schemes, HT-OFE treats candidate pixels almost as foreground until objects complete; a decision is then made to keep or discard these pixels. Processing on the fly enables faster results and avoids additional passes for handling weak pixels and extracting object features. Moreover, labels are reused so only one compact row is buffered and memory requirements are drastically reduced. © 2014 Springer Science+Business Media, LLC.