An Edge Detection Model Based on W-Transform Single Pixel Imaging

To efficiently reduce the required measurement count for single-pixel imaging edge detection, this study devises an edge detection model based on W-transform single-pixel imaging. In this model, the convolution of the DoG (Difference of Gaussians) operator and W-transform basis patterns as modulation patterns are employed to directly extract edge information from the spectral domain of the target object. Different from traditional methods, it achieves the extraction of edge details from an object without imaging. Numerical simulations and experimental results demonstrate that this model extracts edges with higher signal-to-noise ratios compared to the edge extraction using phase-shifted sinusoidal patterns. Additionally, the proposed model reduces the number of modulation patterns required by half, so as to gain double efficiency. The integration of W-transform single-pixel imaging with edge detection offers a novel approach for edge detections without the need for imaging.