TPDFlibImageSimilarityDedupStats

資料型別、影像處理、最佳化

描述

回報 DeduplicateSimilarImages 執行的解碼、帶索引的候選搜尋、取樣確認、可接受誤差、參照改向、酬載節省與選配的物件收集

宣告

TPDFlibImageSimilarityDedupStats = record
  ImagesScanned: Integer;
  ImagesDecoded: Integer;
  DecodeFailures: Integer;
  CanonicalImages: Integer;
  HashCandidates: Integer;
  CandidateComparisons: Integer;
  ImagesDeduplicated: Integer;
  ReferencesRepointed: Integer;
  OrphanObjectsRemoved: Integer;
  MaxAcceptedHashDistance: Integer;
  MaxAcceptedChannelError: Integer;
  SamplesCompared: Int64;
  DuplicatePayloadBytes: Int64;
  MaxAcceptedMeanError: Double;
end;

欄位

ImagesScanned檢驗過的符合條件間接影像串流
ImagesDecoded成功解碼進比較取樣的影像數
DecodeFailuresImages excluded because decoding failed
CanonicalImages接受的匹配分組後保留的不同影像
HashCandidates在語義與距離排除之前列入考慮的帶索引候選
CandidateComparisonsCandidate pairs that reached decoded sample comparison
ImagesDeduplicated被接受為感知重複的影像數
ReferencesRepointed實際執行時被改向的參照,或演練模式中可回報的數量
OrphanObjectsRemoved選配垃圾收集移除的物件
MaxAcceptedHashDistanceLargest difference-hash Hamming distance among accepted pairs
MaxAcceptedChannelError被接受配對中的單一通道最大誤差
SamplesCompared取樣確認檢驗的顏色通道值
DuplicatePayloadBytes被接受的重複影像串流所代表的編碼酬載位元組
MaxAcceptedMeanError被接受配對中的平均絕對通道最大誤差

另請參閱

DeduplicateSimilarImages, GetSimilarImageDeduplicationReportJSON, TPDFlibResourceDedupStats