TIEVisionImageHashAlgorithm
TIEVisionImageHashAlgorithm
Declaration
TIEVisionImageHashAlgorithm = (
ievHashAverage,
ievHashPHash,
ievHashBlockMean,
ievHashMarrHildreth,
ievHashRadialVariance,
ievHashColorMoment
);
Description
The perceptual hash algorithm used by
TIEVisionImageHasher.
A perceptual hash stays (almost) the same when an image is resized, recompressed, slightly blurred or has its brightness changed, so it can be used to find similar and near-duplicate images.
| Value | Hash Size | Description |
| ievHashAverage | 8 bytes | Compares each pixel of an 8x8 gray thumbnail to the mean. Fastest, but the least accurate |
| ievHashPHash | 8 bytes | Uses the low frequencies of a discrete cosine transform. Fast and reliable. This is the recommended algorithm |
| ievHashBlockMean | 32 bytes | Compares the mean of 16x16 blocks to the median. Tolerates small rotations |
| ievHashMarrHildreth | 72 bytes | Uses Marr-Hildreth edge features. Slower, and sensitive to cropping and rotation |
| ievHashRadialVariance | 40 bytes | Uses the variance along radial lines (Radon transform). Tolerates small crops and rotations, but is sensitive to text overlays |
| ievHashColorMoment | 336 bytes | Uses color statistics, so it tolerates crops and rotation, but an image converted to gray or with its brightness changed is not seen as similar |
See Also
◼createImageHasher
◼GetPerceptualHash