Quantum Walk-Based Sliding Window Image Edge Detection Framework
Dibyasree Guha, Nibaran Das
Identifying structural boundaries in an image has emerged as an important image processing and computer vision task rendering edge detection as a fundamental operation. Quantum parallelism and superposition have proven to make positive contributions in such tasks enabling quantum computing to be a foremost paradigm to be adapted for image edge detection purpose. In this work, we propose a generalized quantum-walkbased edge detection framework that operates on overlapping$N \times N$image blocks represented as grid graphs whose vertices correspond to pixel locations. Probability amplitudes associated with edge pixels identified through local gradient criteria, undergo amplification employing a discrete-time coined quantum walk. The dimension-independent formulation improves spatial continuity and reduces block boundary artifacts by probability aggregation producing a continuous edge probability map prior to thresholding. The method is robust to noise and has flexible post-processing capabilities. Comparative analysis with existing quantum walk edge detection methods demonstrates that the proposed framework offers enhanced spatial modeling, improved edge continuity, and greater theoretical generality while retaining the advantages of quantum-walk-based probability amplification. Evaluations of quantitative metrics such as Structural Similarity Index Measures and Pratt's Figure of Merit, further validate the effectiveness of the proposed method.
International Conference on Computing, Intelligence, and Application (CIACON)•Jul 1, 2026•0 citations