{ "__type": "IngestedDoc", "__tag": 4010, "_content": { "Notes": { "__type": "Section", "__tag": 4015, "children": [ { "__type": "Paragraph", "__tag": 4045, "children": [ { "__type": "Text", "__tag": 4046, "value": "A denoising function is J-invariant if the prediction it makes for each pixel does not depend on the value of that pixel in the original image. The prediction for each pixel may instead use all the relevant information contained in the rest of the image, which is typically quite significant. Any function can be converted into a J-invariant one using a simple masking procedure, as described in [1]." } ] }, { "__type": "Paragraph", "__tag": 4045, "children": [ { "__type": "Text", "__tag": 4046, "value": "The pixel-wise error of a J-invariant denoiser is uncorrelated to the noise, so long as the noise in each pixel is independent. Consequently, the average difference between the denoised image and the oisy image, the " }, { "__type": "Emphasis", "__tag": 4047, "children": [ { "__type": "Text", "__tag": 4046, "value": "self-supervised loss" } ] }, { "__type": "Text", "__tag": 4046, "value": ", is the same as the difference between the denoised image and the original clean image, the " }, { "__type": "Emphasis", "__tag": 4047, "children": [ { "__type": "Text", "__tag": 4046, "value": "ground-truth loss" } ] }, { "__type": "Text", "__tag": 4046, "value": " (up to a constant)." } ] }, { "__type": "Paragraph", "__tag": 4045, "children": [ { "__type": "Text", "__tag": 4046, "value": "This means that the best J-invariant denoiser for a given image can be found using the noisy data alone, by selecting the denoiser minimizing the self- supervised loss." } ] } ], "title": [], "level": 0, "target": null }, "Warns": { "__type": "Section", "__tag": 4015, "children": [], "title": [], "level": 0, "target": null }, "Raises": { "__type": "Section", "__tag": 4015, "children": [], "title": [], "level": 0, "target": null }, "Yields": { "__type": "Section", "__tag": 4015, "children": [], "title": [], "level": 0, "target": null }, "Methods": { "__type": "Section", "__tag": 4015, "children": [], "title": [], "level": 0, "target": null }, "Returns": { "__type": "Section", "__tag": 4015, "children": [ { "__type": "Parameters", "__tag": 4026, "children": [ { "__type": "DocParam", "__tag": 4016, "name": "output", "annotation": "ndarray", "desc": [ { "__type": "Paragraph", "__tag": 4045, "children": [ { "__type": "Text", "__tag": 4046, "value": "Denoised image, of same shape as " }, { "__type": "ParamRef", "__tag": 4071, "name": "image" }, { "__type": "Text", "__tag": 4046, "value": "." } ] } ] } ] } ], "title": [], "level": 0, "target": null }, "Summary": { "__type": "Section", "__tag": 4015, "children": [ { "__type": "Paragraph", "__tag": 4045, "children": [ { "__type": "Text", "__tag": 4046, "value": "Apply a J-invariant version of a denoising function." } ] } ], "title": [], "level": 0, "target": null }, "Receives": { "__type": "Section", "__tag": 4015, "children": [], "title": [], "level": 0, "target": null }, "Warnings": { "__type": "Section", "__tag": 4015, "children": [], "title": [], "level": 0, "target": null }, "Attributes": { "__type": "Section", "__tag": 4015, "children": [], "title": [], "level": 0, "target": null }, "Parameters": { "__type": "Section", "__tag": 4015, "children": [ { "__type": "Parameters", "__tag": 4026, "children": [ { "__type": "DocParam", "__tag": 4016, "name": "image", "annotation": "ndarray (M[, N[, ...]][, C]) of ints, uints or floats", "desc": [ { "__type": "Paragraph", "__tag": 4045, "children": [ { "__type": "Text", "__tag": 4046, "value": "Input data to be denoised. " }, { "__type": "ParamRef", "__tag": 4071, "name": "image" }, { "__type": "Text", "__tag": 4046, "value": " can be of any numeric type, but it is cast into a ndarray of floats (using " }, { "__type": "CrossRef", "__tag": 4002, "value": "img_as_float", "reference": { "__type": "LocalRef", "__tag": 4022, "kind": "module", "path": "skimage.util.dtype:img_as_float" }, "kind": "module" }, { "__type": "Text", "__tag": 4046, "value": ") for the computation of the denoised image." } ] } ] }, { "__type": "DocParam", "__tag": 4016, "name": "denoise_function", "annotation": "function", "desc": [ { "__type": "Paragraph", "__tag": 4045, "children": [ { "__type": "Text", "__tag": 4046, "value": "Original denoising function." } ] } ] }, { "__type": "DocParam", "__tag": 4016, "name": "stride", "annotation": "int, optional", "desc": [ { "__type": "Paragraph", "__tag": 4045, "children": [ { "__type": "Text", "__tag": 4046, "value": "Stride used in masking procedure that converts " }, { "__type": "ParamRef", "__tag": 4071, "name": "denoise_function" }, { "__type": "Text", "__tag": 4046, "value": " to J-invariance." } ] } ] }, { "__type": "DocParam", "__tag": 4016, "name": "masks", "annotation": "list of ndarray, optional", "desc": [ { "__type": "Paragraph", "__tag": 4045, "children": [ { "__type": "Text", "__tag": 4046, "value": "Set of masks to use for computing J-invariant output. If " }, { "__type": "InlineRole", "__tag": 4003, "value": "None", "domain": null, "role": null, "inventory": null }, { "__type": "Text", "__tag": 4046, "value": ", a full set of masks covering the image will be used." } ] } ] }, { "__type": "DocParam", "__tag": 4016, "name": "denoiser_kwargs", "annotation": "", "desc": [ { "__type": "Paragraph", "__tag": 4045, "children": [ { "__type": "Text", "__tag": 4046, "value": "Keyword arguments passed to " }, { "__type": "ParamRef", "__tag": 4071, "name": "denoise_function" }, { "__type": "Text", "__tag": 4046, "value": "." } ] } ] } ] } ], "title": [], "level": 0, "target": null }, "Extended Summary": { "__type": "Section", "__tag": 4015, "children": [], "title": [], "level": 0, "target": null }, "Other Parameters": { "__type": "Section", "__tag": 4015, "children": [], "title": [], "level": 0, "target": null } }, "_ordered_sections": [ "Summary", "Extended Summary", "Parameters", "Returns", "Yields", "Receives", "Raises", "Warns", "Other Parameters", "Attributes", "Methods", "Notes", "Warnings" ], "item_file": "/dev/scikit-image/src/skimage/restoration/j_invariant.py", "item_line": 90, "item_type": "function", "aliases": [ "skimage.restoration.denoise_invariant" ], "example_section_data": { "__type": "Section", "__tag": 4015, "children": [ { "__type": "Code", "__tag": 4050, "value": "import skimage\nfrom skimage.restoration import denoise_invariant, denoise_tv_chambolle\nimage = skimage.util.img_as_float(skimage.data.chelsea())\nnoisy = skimage.util.random_noise(image, var=0.2 ** 2)\n", "execution_status": "success" } ], "title": [], "level": 0, "target": null }, "see_also": [], "signature": { "__type": "SignatureNode", "__tag": 4029, "kind": "function", "parameters": [ { "__type": "SigParam", "__tag": 4030, "name": "image", "annotation": { "__type": "Empty", "__tag": 4031 }, "kind": "POSITIONAL_OR_KEYWORD", "default": { "__type": "Empty", "__tag": 4031 } }, { "__type": "SigParam", "__tag": 4030, "name": "denoise_function", "annotation": { "__type": "Empty", "__tag": 4031 }, "kind": "POSITIONAL_OR_KEYWORD", "default": { "__type": "Empty", "__tag": 4031 } }, { "__type": "SigParam", "__tag": 4030, "name": "stride", "annotation": { "__type": "Empty", "__tag": 4031 }, "kind": "KEYWORD_ONLY", "default": "4" }, { "__type": "SigParam", "__tag": 4030, "name": "masks", "annotation": { "__type": "Empty", "__tag": 4031 }, "kind": "KEYWORD_ONLY", "default": "None" }, { "__type": "SigParam", "__tag": 4030, "name": "denoiser_kwargs", "annotation": { "__type": "Empty", "__tag": 4031 }, "kind": "KEYWORD_ONLY", "default": "None" } ], "return_annotation": { "__type": "Empty", "__tag": 4031 }, "target_name": "denoise_invariant" }, "references": [ ".. [1] J. Batson & L. Royer. Noise2Self: Blind Denoising by Self-Supervision,", " International Conference on Machine Learning, p. 524-533 (2019)." ], "qa": "skimage.restoration.j_invariant:denoise_invariant", "arbitrary": [], "local_refs": [ "denoise_function", "denoiser_kwargs", "image", "masks", "output", "stride" ] }