TIEVisionFeaturesDetector.setMaxImageSize

Declaration

procedure setMaxImageSize(megapixels: Double); safecall;

Description

Images larger than megapixels are reduced before keypoints are detected. The keypoints and matching points are still returned in the coordinates of the original images.
This limits memory use and time: SIFT needs about 250 bytes and KAZE about 550 bytes of memory per pixel, so detecting keypoints in a 24 MP photo at full size would need 6 to 13 GB. Alignment is generally as accurate (or more accurate) at the reduced size.
0 means no limit.

Default:

Example

// Use full resolution for more precise keypoint positions in small images
detector := IEVisionLib().createFeaturesDetector( ievFeatureSIFT );
detector.setMaxImageSize( 0 );

See Also

◼getMaxImageSize
◼detect

---