引導式測驗 · 中等 等級

近似相同圖像的感知雜湊測驗

測試視覺與加密哈希、漢明距離、閾值權衡、裁剪和壓縮限制以及候選匹配的審查。

問題 1 的 8

Two JPEG files look alike but differ in bytes. Why can their cryptographic hashes differ?

預覽每個問題8 問題

當您參加測驗時,就會出現答案和解釋。

  1. Two JPEG files look alike but differ in bytes. Why can their cryptographic hashes differ?
    • Visual similarity requires byte identity
    • Cryptographic digests are designed for exact byte integrity
    • JPEG compression cannot alter bytes
    • A cryptographic digest compares semantic content
  2. What does Hamming distance count for two binary perceptual hashes?
    • The number of detected objects
    • The pixel-width difference
    • The difference in compressed file sizes
    • The bit positions where their signatures differ
  3. Why validate a near-duplicate distance threshold on the target collection?
    • It trades false matches against missed similar images
    • One distance threshold fits every hash algorithm
    • A threshold identifies the original rights holder
    • Distance remains unchanged after every crop
  4. Why might a large crop defeat a perceptual hash that tolerates mild compression?
    • Cropping only changes encoding noise
    • Perceptual hashes compare file bytes exactly
    • Cropping changes global visual structure rather than only encoding noise
    • Any algorithm must ignore composition changes
  5. A library finds two close hashes. What action is safest before deleting one file?
    • Low distance proves the photos are identical
    • Visually review the candidates and keep an undo path
    • The older filename establishes ownership
    • A similar signature guarantees one product
  6. Which purpose fits a cryptographic digest better than a perceptual hash?
    • Checking whether a stored file’s bytes changed
    • Finding a recompressed copy visually
    • Grouping versions after a large crop
    • Matching object style across photos
  7. Two similar product photos have close perceptual hashes. Which claim remains unsupported?
    • They show the same exact item and seller
    • Their signatures are close under this method
    • They are candidates for human review
    • They show the exact same item from the same seller
  8. Why should an audit record the hash method and threshold?
    • All methods produce the same signature
    • Recording settings establishes authorship
    • A distance result depends on algorithm and decision setting
    • Thresholds are determined by JPEG format alone