Unverified50% confidenceFactExact time
PhotoDNA使用感知哈希（perceptual hash），通过灰度转换、缩放、DCT系数提取和汉明距离比较实现对JPEG压缩、轻微裁剪和色调调整的鲁棒性
1
Sources
50%
Confidence
Long-term
Relevance
7/9/2026
First Seen
Sources
Related Claims
UnverifiedPhotoDNA基于鲁棒性哈希值（perceptual hash），即便图像经过裁剪、压缩或色彩调整哈希值仍保持稳定，但只能识别已知违规内容及其变体83% similarUnverifiedPhotoDNA converts images into digital fingerprints that can identify known illegal content even after the image has been cropped or color-adjusted, and the matching process does not require human review of the original image.77% similarUnverified感知哈希对视觉相似图像输出相近哈希值，与密码学哈希（如SHA-256）的雪崩效应有本质区别，能识别经裁剪、压缩或色彩调整的图片63% similarUnverifiedPhotorealistic portraits generated by GPT-Image-2 closely approximate reality in terms of skin texture, light-and-shadow relationships, and subtle facial expressions, to the point where even users with three years of AI image generation experience find it difficult to distinguish them from real photos.56% similarUnverified研究者通过分析图像频域特征（傅里叶变换下的频率分布异常）、面部生物力学一致性及全局光照方向来识别AI合成图像56% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/365364API
curl https://kongchang.com/api/v1/knowledge/claims/365364MCP
get_claim(id=365364)