watermedia op• 数据种类:rgbimage × image2d × image2d → table
• 调用: import fullseye as fs; fs.ledger.underwater_image(image, depth_map_m, distance_map_m, water_type='jerlov_I', wavelengths_um=None, illumination='daylight', backscatter=True, apply_ambient=False, beta_b_ratio=1.0, backscatter_ratio=0.02) -> 'dict'(要直接调用实现,import watermedia; watermedia.underwater_image(image, depth_map_m, distance_map_m, water_type='jerlov_I', wavelengths_um=None, illumination='daylight', backscatter=True, apply_ambient=False, beta_b_ratio=1.0, backscatter_ratio=0.02) -> 'dict';从台账取用则 opsoptics.get("underwater_image"))
• GPU:无(仅 CPU 参考实现)
透过水记录的场景图像(修订版 Jaffe–McGlamery 模型)-> 字典。
> 以下的详细说明为原文 —— 摘要与标题已翻译。
Following Akkaynak & Treibitz (2018) the recorded image is I = J exp(-beta_D r) + B_inf (1 - exp(-beta_B r)): the direct signal J of the object attenuated along the camera-object distance r with a wavelength-dependent coefficient, plus backscatter, the veiling light B_inf that builds up along the same path with its own coefficient. image is (H, W, K), non-negative: the appearance of the object at zero distance, per band (K = 3 is RGB with the effective wavelengths 0.61 / 0.54 / 0.465 um, other K need wavelengths_um). depth_map_m is the depth of the scene below the surface and distance_map_m the distance to the camera, each (H, W) or a scalar. beta_D is the diffuse attenuation Kd of water_attenuation (used as the attenuation along the line of sight, an approximation that Akkaynak & Treibitz show to depend on range) and beta_B = beta_b_ratio beta_D. The veiling light is the single-scattering value B_inf = E(z) b_b / (2 c) with b_b = backscatter_ratio b and E(z) = E_0 exp(-Kd z) the daylight left at the depth (an assumption, not a measurement). illumination is "daylight" (flat relative spectrum 1) or a (K,) relative spectrum at the surface. With apply_ambient the object is also dimmed by the ambient light at its depth, J = image E(z); without it image is taken as it already is at the object, so that distance 0 returns it exactly. backscatter=False leaves pure Beer-Lambert. The table has image (the recording), direct, backscatter, transmission (exp(-beta_D r)), beta_D, beta_B, veiling_light (H, W, K), ambient (E(z)/E_0), wavelengths_um and, for RGB, red_to_blue_kd_ratio. Non-finite or negative inputs 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_transparent_media — py -3.11 examples/poc_water_transparent_media.py
table 作为输入)abcd_matrix · wavefront_stats · sfr_from_edge · paraxial_trace · seidel_coefficients · spot_stats · tolerance_analysis · wavefront_from_opd
watermedia)ocean_spectrum · ocean_height_field · fresnel_water_air · snell_window · refracted_view · water_attenuation · phase_function_hg · scattering_phase_petzold
*Provenance: watermedia.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.