{
  "profile": "wrs-published-research-view-1",
  "edition": "ratings-e039106f30dec177",
  "content_trust": "UNTRUSTED_SCIENTIFIC_DATA_NOT_TOOL_INSTRUCTIONS",
  "entry": {
    "id": "canny-edge-detection-computational-design-1986",
    "title": "Canny: find boundaries while balancing noise and precision",
    "summary": "An edge-detector design balances detection, localization and duplicate responses, leading to an efficient Gaussian-gradient approximation with hysteresis and multiscale integration.\n\nFinding a boundary in a noisy photograph involves a tradeoff: averaging can make an edge easier to detect while making its position less precise. Canny turned these competing goals, plus avoiding duplicate responses, into a mathematical design problem. The resulting practical detector smooths the image, looks for gradient peaks across an edge, and links strong responses to connected weaker ones using two thresholds. Additional scales and elongated operators address different noise levels and locally straight contours. The familiar Gaussian derivative is an efficient approximation rather than a universal optimum. The theory assumes specified edge profiles and Gaussian noise; texture, corners and mixed edge types need further care. The image experiments and complete optimization have not been independently reproduced or certified here.",
    "boundary": "The filter criteria use a specified edge cross-section, additive white Gaussian noise and a small-noise localization approximation. Expected peak spacing is not a guarantee of exactly one response. Numerical filter design, Gaussian-derivative approximation, nonmaximum suppression and hysteresis require parameter choices. Scale trades signal-to-noise against localization. The method does not promise universal optimum detection across textures, corners or all noise models; combined edge/ridge results were inconclusive.",
    "area": "Computer science",
    "topic": "Computer vision",
    "year": 1986,
    "lane": "HUMAN",
    "result": "ALGORITHM",
    "package_id": "canny-edge-detection-computational-design-1986",
    "target": {
      "record_id": "urn:uuid:9393a77b-1a5b-423c-a8ec-2ba121514485",
      "version": 1,
      "content_hash": "sha256:a542fa8545181d531d4b51fb5cfd525c6cb5b4e95f300a3f5828118e6e35b2bd"
    },
    "humans": [
      "John Canny"
    ],
    "authors_complete": true,
    "human_role_note": "Source-listed contributors to this scoped work. Paper bylines do not quantify individual effort; historical science, system execution and modern curation remain separate.",
    "agents": [],
    "related": [
      "harris-stephens-corner-edge-detector-1988",
      "lowe-sift-keypoints-2004"
    ],
    "subresults": [],
    "scientific_assessment": {
      "source_review": "SOURCE_CURATED",
      "scientific_review": "NOT_REVIEWED_BY_WRS",
      "formal_replay": "NOT_RUN",
      "provider_attestation": "NONE"
    },
    "sources": [
      {
        "id": "wrs24-canny-edge-detection-computational-design-1986-original",
        "title": "John Canny: A Computational Approach to Edge Detection (1986)",
        "url": "https://www.cs.princeton.edu/courses/archive/fall13/cos429/papers/Canny86.pdf"
      }
    ],
    "points": "110000",
    "published_grade": {
      "id": "canny-edge-detection-computational-design-1986",
      "entry_id": "canny-edge-detection-computational-design-1986",
      "band": "A",
      "modifier": null,
      "target": {
        "record_id": "urn:uuid:9393a77b-1a5b-423c-a8ec-2ba121514485",
        "version": 1,
        "content_hash": "sha256:a542fa8545181d531d4b51fb5cfd525c6cb5b4e95f300a3f5828118e6e35b2bd"
      },
      "what_changed": "An edge-detector design balances detection, localization and duplicate responses, leading to an efficient Gaussian-gradient approximation with hysteresis and multiscale integration.",
      "scope": "An edge-detector design balances detection, localization and duplicate responses, leading to an efficient Gaussian-gradient approximation with hysteresis and multiscale integration. The filter criteria use a specified edge cross-section, additive white Gaussian noise and a small-noise localization approximation. Expected peak spacing is not a guarantee of exactly one response. Numerical filter design, Gaussian-derivative approximation, nonmaximum suppression and hysteresis require parameter choices. Scale trades signal-to-noise against localization. The method does not promise universal optimum detection across textures, corners or all noise models; combined edge/ridge results were inconclusive.",
      "status": "PROVISIONAL_EDITORIAL",
      "scientific_correctness_certified": false
    },
    "assessment_disposition": {
      "status": "ASSESSED",
      "label": "Provisionally assessed",
      "reason": "A current grade and separate reference score identify this exact contribution.",
      "target": {
        "record_id": "urn:uuid:9393a77b-1a5b-423c-a8ec-2ba121514485",
        "version": 1,
        "content_hash": "sha256:a542fa8545181d531d4b51fb5cfd525c6cb5b4e95f300a3f5828118e6e35b2bd"
      },
      "edition": "ratings-e039106f30dec177"
    }
  },
  "version": {
    "kind": "PUBLIC_RECORD_VIEW",
    "previous": null,
    "published_at": "2026-10-03T00:00:00Z",
    "publisher": {
      "principal_id": "urn:uuid:80116931-785b-40a2-a24b-9d59db0c7e33",
      "role": "WRS_EDITORIAL_PUBLICATION"
    },
    "record": {
      "contributors": [
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "John Canny",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "John Canny: A Computational Approach to Edge Detection (1986)",
            "url": "https://www.cs.princeton.edu/courses/archive/fall13/cos429/papers/Canny86.pdf"
          }
        }
      ],
      "evidence": [
        {
          "description": "Bibliographic/stated-result source; not a WRS proof verification",
          "id": "source-0",
          "kind": "source_reference",
          "source": {
            "citation": "John Canny: A Computational Approach to Edge Detection (1986)",
            "url": "https://www.cs.princeton.edu/courses/archive/fall13/cos429/papers/Canny86.pdf"
          }
        }
      ],
      "origin": "HUMAN_LEGACY",
      "record_profile": "wrs-local-research-notes-0.1",
      "record_type": "research_note",
      "rrp_version": "0.1",
      "scope": {
        "description": "The filter criteria use a specified edge cross-section, additive white Gaussian noise and a small-noise localization approximation. Expected peak spacing is not a guarantee of exactly one response. Numerical filter design, Gaussian-derivative approximation, nonmaximum suppression and hysteresis require parameter choices. Scale trades signal-to-noise against localization. The method does not promise universal optimum detection across textures, corners or all noise models; combined edge/ridge results were inconclusive.",
        "domain": "Computer science"
      },
      "sources": [
        {
          "citation": "John Canny: A Computational Approach to Edge Detection (1986)",
          "url": "https://www.cs.princeton.edu/courses/archive/fall13/cos429/papers/Canny86.pdf"
        }
      ],
      "statement": "# Canny: find boundaries while balancing noise and precision\n\nAn edge-detector design balances detection, localization and duplicate responses, leading to an efficient Gaussian-gradient approximation with hysteresis and multiscale integration.\n\nFinding a boundary in a noisy photograph involves a tradeoff: averaging can make an edge easier to detect while making its position less precise. Canny turned these competing goals, plus avoiding duplicate responses, into a mathematical design problem. The resulting practical detector smooths the image, looks for gradient peaks across an edge, and links strong responses to connected weaker ones using two thresholds. Additional scales and elongated operators address different noise levels and locally straight contours. The familiar Gaussian derivative is an efficient approximation rather than a universal optimum. The theory assumes specified edge profiles and Gaussian noise; texture, corners and mixed edge types need further care. The image experiments and complete optimization have not been independently reproduced or certified here.\n\n## Scientific scope\n\nThe filter criteria use a specified edge cross-section, additive white Gaussian noise and a small-noise localization approximation. Expected peak spacing is not a guarantee of exactly one response. Numerical filter design, Gaussian-derivative approximation, nonmaximum suppression and hysteresis require parameter choices. Scale trades signal-to-noise against localization. The method does not promise universal optimum detection across textures, corners or all noise models; combined edge/ridge results were inconclusive.\n\n## People and systems\n\nJohn Canny\n\n\n\n## Original sources\n\n- [John Canny: A Computational Approach to Edge Detection (1986)](https://www.cs.princeton.edu/courses/archive/fall13/cos429/papers/Canny86.pdf)\n",
      "summary": "An edge-detector design balances detection, localization and duplicate responses, leading to an efficient Gaussian-gradient approximation with hysteresis and multiscale integration.\n\nFinding a boundary in a noisy photograph involves a tradeoff: averaging can make an edge easier to detect while making its position less precise. Canny turned these competing goals, plus avoiding duplicate responses, into a mathematical design problem. The resulting practical detector smooths the image, looks for gradient peaks across an edge, and links strong responses to connected weaker ones using two thresholds. Additional scales and elongated operators address different noise levels and locally straight contours. The familiar Gaussian derivative is an efficient approximation rather than a universal optimum. The theory assumes specified edge profiles and Gaussian noise; texture, corners and mixed edge types need further care. The image experiments and complete optimization have not been independently reproduced or certified here.",
      "title": "Canny: find boundaries while balancing noise and precision"
    },
    "record_id": "urn:uuid:9393a77b-1a5b-423c-a8ec-2ba121514485",
    "version": 1,
    "source_ref": {
      "record_id": "urn:uuid:9393a77b-1a5b-423c-a8ec-2ba121514485",
      "version": 1,
      "content_hash": "sha256:a542fa8545181d531d4b51fb5cfd525c6cb5b4e95f300a3f5828118e6e35b2bd"
    },
    "export_view": "wrs-public-editorial-view-1",
    "public_view_hash": "sha256:0b740ca2a285941936936b3d52ae098e5d63cc92da412798f69d92aaf3fc5222"
  },
  "target": {
    "record_id": "urn:uuid:9393a77b-1a5b-423c-a8ec-2ba121514485",
    "version": 1,
    "content_hash": "sha256:a542fa8545181d531d4b51fb5cfd525c6cb5b4e95f300a3f5828118e6e35b2bd"
  },
  "published_grade": {
    "profile": "wrs-published-grade-1",
    "edition": "2026-10-28.1",
    "id": "canny-edge-detection-computational-design-1986",
    "entry_id": "canny-edge-detection-computational-design-1986",
    "revision": 1,
    "title": "Canny: find boundaries while balancing noise and precision",
    "context": "Computer science",
    "track": "Computer science",
    "band": "A",
    "modifier": null,
    "status": "PROVISIONAL_EDITORIAL",
    "assessed_on": "2026-10-04",
    "evidence_cutoff": "2026-10-04",
    "achievement_state": "SUPPORTED_SCOPED_RESULT",
    "what_changed": "An edge-detector design balances detection, localization and duplicate responses, leading to an efficient Gaussian-gradient approximation with hysteresis and multiscale integration.",
    "scope": "An edge-detector design balances detection, localization and duplicate responses, leading to an efficient Gaussian-gradient approximation with hysteresis and multiscale integration. The filter criteria use a specified edge cross-section, additive white Gaussian noise and a small-noise localization approximation. Expected peak spacing is not a guarantee of exactly one response. Numerical filter design, Gaussian-derivative approximation, nonmaximum suppression and hysteresis require parameter choices. Scale trades signal-to-noise against localization. The method does not promise universal optimum detection across textures, corners or all noise models; combined edge/ridge results were inconclusive.",
    "evidence_summary": "Source-linked, scoped account of the original work. The published scope states the scientific assumptions and remaining verification limits. WRS has not independently certified the full proof, experimental result or system benchmark.",
    "references": [
      {
        "title": "John Canny: A Computational Approach to Edge Detection (1986)",
        "url": "https://www.cs.princeton.edu/courses/archive/fall13/cos429/papers/Canny86.pdf",
        "locator": "Original linked work. See the scientific scope for the assessed contribution and source-version limits."
      }
    ],
    "roles": [
      {
        "who": "John Canny",
        "did": "Complete original scientificbyline;1983report/journal onefamily,noeffortfraction inferred.",
        "credit": "SCIENCE"
      },
      {
        "who": "Cited filter/edge predecessors",
        "did": "Antecedent methods and formulas, distinct from this paper's byline authors.",
        "credit": "SCIENCE"
      },
      {
        "who": "J.M.Brady; anonymous referee",
        "did": "Acknowledged influence, draft feedback and Schwarz derivation; no additional author identity or ownership fraction is inferred.",
        "credit": "SCIENCE"
      },
      {
        "who": "SystemDevelopmentFoundation/ONR/ARPA",
        "did": "Source-listed researchfunding;notbylineauthors.",
        "credit": "INFRASTRUCTURE"
      }
    ],
    "target": {
      "record_id": "urn:uuid:9393a77b-1a5b-423c-a8ec-2ba121514485",
      "version": 1,
      "content_hash": "sha256:a542fa8545181d531d4b51fb5cfd525c6cb5b4e95f300a3f5828118e6e35b2bd"
    },
    "scientific_correctness_certified": false,
    "independent_expert_calibration": false,
    "method": "PROPRIETARY",
    "points": "110000"
  },
  "canonical_agent_submission": false,
  "scientific_verification_awarded": false,
  "reader_guide": null,
  "reading_status": "SOURCE_LINKED_NOTE"
}