Object density layer edge detection with adaptive anisotropy factors and multiscale fusion

Aiming at the problem of edge extraction for reconstruction results of high energy flash X-ray images, an edge detection algorithm for each density layer of layer based on adaptive anisotropy factor and multi-scale fusion is proposed. The anisotropy factor, which is associated with numbers of anisotropic filters, is derived based on the ratio of signal-to-noise ratio of the edge strength to the resolution constant. The adaptive anisotropy factor can reduce anisotropic stretch phenomenon and spurious edge response. Edge intensity maps between adjacent scales are fused by scale multiplication to reduce the conflict between high edge resolution and robust to noise so as to enhance edge structure. The edge strength maps obtained under the first-order derivative of anisotropic and isotropic Gaussian kernels are further fused by geometric averaging mechanism. Finally, the false edge suppression of non-maximum and double threshold is carried out to obtain the edge with single-pixel of the target. Experimental results show that the algorithm can effectively suppress the influence of noise and reduce the fuzzy effect on edge, and can obtain more accurate edge detection results.