Resource-Aware Harris Corner Detection Based on Adaptive Pruning

摘要: Corner-detection techniques are being widely used in computer vision — for example object recognition to find suitable candidate points feature registration and matching. Most computer-vision applications have operate on real-time video sequences, hence maintaining a consistent throughput high accuracy important constrains that ensure high-quality recognition. A can be achieved by exploiting the inherent parallelism within algorithm massively parallel architectures like many-core processors. However, accelerating such algorithms CPUs offers several challenges as speedup depends instantaneous load processing elements. In this work, we present new resource-aware Harris corner-detection The novel adapt itself dynamically varying processor process frame predefined time interval. results show 19% improvement an 18% accuracy.