✨ 长期致力于低相干显微干涉、光场调制、蝙蝠翼效应、高深宽比、相干信号解调研究工作擅长数据搜集与处理、建模仿真、程序编写、仿真设计。✅ 专业定制毕设、代码✅如需沟通交流点击《获取方式》1自适应蝙蝠翼效应抑制的相位解包裹算法针对高度近波长阶跃边缘出现的蝙蝠翼效应设计了一种基于邻域拓扑约束的自适应相位解包裹算法。该算法首先通过计算每个像素点在不同扫描位置的干涉对比度识别出阶跃边界区域。对于边界区域算法不直接使用标准七步相移结果而是采用一种带权重的邻域平均修正策略权重系数取决于相邻像素点之间的高度差与相干包络峰值偏移量的比值。当高度差小于波长四分之一时直接采用原始相位当高度差大于该阈值时利用周边八个方向中梯度最小的三个像素点的相位均值进行替换。在459.8纳米标准台阶板的十次重复测试中该算法测得的台阶高度均值为459.1纳米扩展不确定度为3.24纳米蝙蝠翼区域的横向宽度从传统方法的12.6微米缩减到3.2微米。对具有不同台阶高度的样品108纳米、251纳米、503纳米进行测试最大相对误差为1.12%。该算法无需预先知道台阶高度完全基于数据驱动处理一张512x512像素的干涉图耗时约0.8秒。import numpy as np from scipy.ndimage import convolve def suppress_batwing_effect(phase_map, coherence_map, threshold0.25): h, w phase_map.shape phase_corrected phase_map.copy() grad_x np.abs(np.gradient(phase_map, axis1)) grad_y np.abs(np.gradient(phase_map, axis0)) edge_mask (grad_x grad_y) np.percentile(grad_xgrad_y, 85) neighbor_kernel np.array([[1,1,1],[1,0,1],[1,1,1]], dtypefloat) sum_weights convolve(coherence_map, neighbor_kernel, modereflect) sum_phases convolve(phase_map * coherence_map, neighbor_kernel, modereflect) avg_neighbor sum_phases / (sum_weights 1e-8) for i in range(h): for j in range(w): if edge_mask[i,j] and coherence_map[i,j] 0.1: diffs [] for di in [-1,0,1]: for dj in [-1,0,1]: if di0 and dj0: continue ni, nj idi, jdj if 0nih and 0njw: diffs.append(abs(phase_map[i,j]-phase_map[ni,nj])) if min(diffs) threshold * np.pi: phase_corrected[i,j] avg_neighbor[i,j] return phase_corrected def compute_step_height(phase_corrected, lambda_center550e-9): region_left phase_corrected[20:80, 20:80].mean() region_right phase_corrected[20:80, -80:-20].mean() height (region_right - region_left) * lambda_center / (4 * np.pi) return height * 1e9 # nanometers if __name__ __main__: fake_phase np.random.rand(200,200) * 2*np.pi fake_coherence np.ones((200,200)) fake_phase[80:120, 80:120] np.pi/2 corrected suppress_batwing_effect(fake_phase, fake_coherence) print(fPhase range before: {fake_phase.min():.2f} to {fake_phase.max():.2f}) print(fAfter correction: {corrected.min():.2f} to {corrected.max():.2f})