watervision op• 数据种类:image2d × image2d → table
• 调用: import fullseye as fs; fs.ledger.wet_dry_inspection_pair(texture, defect_mask, roughness, water_film_um, light_deg, view_deg, n_water=None, seed=268, normal_sigma_px=2.0) -> 'dict'(要直接调用实现,import watervision; watervision.wet_dry_inspection_pair(texture, defect_mask, roughness, water_film_um, light_deg, view_deg, n_water=None, seed=268, normal_sigma_px=2.0) -> 'dict';从台账取用则 opsoptics.get("wet_dry_inspection_pair"))
• GPU:无(仅 CPU 参考实现)
分别渲染同一块有纹理的板的干燥与湿润状态,并给出各自缺陷的对比度 -> 字典。
> 以下的详细说明为原文 —— 摘要与标题已翻译。
texture (H, W) in [0, 1] is the diffuse albedo of the plate and defect_mask (H, W, bool) marks the defect (at least one pixel inside and one outside). Both images use the per-pixel wet_surface_appearance for the diffuse part. The specular part uses a local normal: a Gaussian-smoothed random height map (seed, smoothing normal_sigma_px) whose slopes are scaled to an RMS slope per axis of roughness; each pixel mirrors the light with the Fresnel reflectance of its local incidence angle into the Gaussian lobe of wet_surface_appearance about the viewing direction, so a glint appears where the local normal bisects the light and the view and moves when light_deg changes. The dry and the wet plate share the normals, so they differ only by the film. light_deg is in the xz plane from the plate normal, the viewer on the other side at view_deg. The defect contrast is the Michelson contrast (mean_ok - mean_defect) / (mean_ok + mean_defect) of the region means, reported for the whole image and for the diffuse part alone. Because the film darkens dark pixels proportionally more (the factor (1 - r) / (1 - a r) grows with a), a dark defect on a brighter plate gains diffuse contrast when wet; the glints (shared by both regions) pull the total contrast the other way, so the table has both and predicted_diffuse_contrast_wet (the one-step prediction from the region mean albedos).
The table has dry, wet, diffuse_dry, diffuse_wet, specular_dry, specular_wet (images), contrast_dry, contrast_wet, contrast_diffuse_dry, contrast_diffuse_wet, predicted_diffuse_contrast_wet, coverage, glint_pixels_dry / glint_pixels_wet (pixels whose specular term exceeds the median diffuse albedo) and normals. A film of 0 gives identical images. Shapes that differ, a mask without both classes, textures outside [0, 1] or a negative film are refused.
optics 的每个算子都先校验输入再计算(不让任何东西无声通过):
• 单位写进参数名 —— _mm / _um / _deg / _mrad。把 mm 和 µm 弄混不会崩溃,而是给出「看着合理却是错的答案」,所以用命名来防。这里绝不从数值大小去猜单位。
• 字符串一律 ValueError —— float('50') 会成功,于是未解析的配置值会被当成长度混进来(实测:thin_lens('50', '200') 曾返回看着合理的 66.667 mm)。bool 也按 True == 1 的隐式提升拒绝。
• complex / masked array 一律 ValueError(仅接受实数槽位;拒绝无声丢弃虚部或剥掉掩码)。所有输入中的 NaN/Inf 一律 ValueError。
• 逐项点名拒绝除零及其近亲:焦距 0、曲率半径 0、折射率 <= 0、全不透明光阑(全为 0,归一化变成 0/0)、总和 <= 0 的 PSF、S0 = 0 的 Stokes 矢量、物体位于前焦点(像在无穷远)。
• 只有两个算子会返回非有限值,而且都写进了契约:depth_of_field 在超焦距以外返回 far_mm = inf(这正是超焦距的定义),gaussian_beam 在束腰处返回 wavefront_radius_mm = inf(平面波前的曲率半径)。两者都同时返回一个有限的搭档(far_is_infinite / curvature_per_mm)。除此之外的无声 NaN/Inf 都在内部检出并 ValueError ——「float64 溢出了」和「答案是无穷大」是两种不同的主张,不能拿后者的脸去交付前者。
• 尺寸上限:生成网格受 optics.MAX_GRID(4096)限制,传入的场/PSF/光阑受 optics.MAX_FIELD_ELEMENTS(2^24),ABCD 元件序列受 optics.MAX_SYSTEM_ELEMENTS(1024),Zernike 受 MAX_ZERNIKE_TERMS(512)/ MAX_ZERNIKE_ORDER(40)/ MAX_ZERNIKE_BASIS(2^25)。以 fail-closed 堵住「小参数引发巨大内部分配」的路径(实测:n_max=40 × 4096² 需要 108 GB)。
• 物理上不可能的状态同样拒绝:偏振度 > 1 的 Stokes 矢量、负透过率、负强度、n-|m| 为奇数等非法 Zernike 指标。
• 示例数据目录(下载 URL / 许可证) —— 2-D 用 skimage.data(BSD/公有领域)加合成图,3-D 给出真实数据源(Stanford/PDS 等)的下载 URL。
• 算子来历与参考文献 —— 该算子族所依据的研究/方法出处。
• 算法的正典(作者、年份)与用途见上面的族使用指南。
• poc_water_inspection_images — py -3.11 examples/poc_water_inspection_images.py
table 作为输入)abcd_matrix · wavefront_stats · sfr_from_edge · paraxial_trace · seidel_coefficients · spot_stats · tolerance_analysis · wavefront_from_opd
watervision)caustic_irradiance · bottle_image · fill_level_from_image · wet_surface_appearance · underwater_inspection_image · caustic_floor_segmentation
*Provenance: watervision.py — OPTICS 算子登记表。本条目由 tools/opdocs.py md 自动生成(请勿手工编辑)。*
© 2026 Kazufumi Furuse — Fullseye operator documentation. Licensed under Apache-2.0.