philip-lamb
commented
6 years ago I believe the adjustment to the initializing features is no longer supported as of ARToolKit v5.3. It was a feature which adjusted the SURF feature extraction sensitivity in ARToolKit v5.2. There may be a way to adjust the sensitivity of the FREAK detector in 5.3 but I haven't looked into it (I didn't write that code).
I run gentexdata on the same image twice, once on the original image and once on a copy of the image. The first time i select 4, which is the highest value for extraction of tracking features, 3 the highest for initializing features and 10 for mindpi 80 for maxdpi. The second time i select 0 (lowest) for tracking features and 0 (lowest) for initializing features, mindpi 10 and maxdpi 80.
when i run dispfeatureset on the two fsets i get the following:
Extraction level for tracking features 4 Extraction level for initializing features 3 Min_dpi 10 Max_dpi 80
Size = (1000,563) Zoom = 1.000000 80.000000[dpi] image. Size = (1000,563) fset: Num of feature points: 222 fset3: Num of feature points: 340 fset: Num of feature points: 222 fset3: Num of feature points: 340 63.496048[dpi] image. Size = (794,447) fset: Num of feature points: 155 fset3: Num of feature points: 323 50.396847[dpi] image. Size = (630,355) fset: Num of feature points: 90 fset3: Num of feature points: 618 40.000004[dpi] image. Size = (500,282) fset: Num of feature points: 61 fset3: Num of feature points: 408 31.748024[dpi] image. Size = (397,223) fset: Num of feature points: 34 fset3: Num of feature points: 250 25.198423[dpi] image. Size = (315,177) fset: Num of feature points: 26 fset3: Num of feature points: 194 20.000002[dpi] image. Size = (250,141) fset: Num of feature points: 13 fset3: Num of feature points: 143 15.874011[dpi] image. Size = (198,112) fset: Num of feature points: 9 fset3: Num of feature points: 90 12.599211[dpi] image. Size = (157,89) fset: Num of feature points: 5 fset3: Num of feature points: 52 10.000000[dpi] image. Size = (125,70) fset: Num of feature points: 3 fset3: Num of feature points: 41
Extraction level for tracking features 0 Extraction level for initializing features 0 Min_dpi 10 Max_dpi 80
Size = (1000,563) Zoom = 1.000000 80.000000[dpi] image. Size = (1000,563) fset: Num of feature points: 28 fset3: Num of feature points: 340 fset: Num of feature points: 28 fset3: Num of feature points: 340 63.496048[dpi] image. Size = (794,447) fset: Num of feature points: 18 fset3: Num of feature points: 323 50.396847[dpi] image. Size = (630,355) fset: Num of feature points: 19 fset3: Num of feature points: 618 40.000004[dpi] image. Size = (500,282) fset: Num of feature points: 9 fset3: Num of feature points: 408 31.748024[dpi] image. Size = (397,223) fset: Num of feature points: 5 fset3: Num of feature points: 250 25.198423[dpi] image. Size = (315,177) fset: Num of feature points: 4 fset3: Num of feature points: 194 20.000002[dpi] image. Size = (250,141) fset: Num of feature points: 3 fset3: Num of feature points: 143 15.874011[dpi] image. Size = (198,112) fset: Num of feature points: 2 fset3: Num of feature points: 90 12.599211[dpi] image. Size = (157,89) fset: Num of feature points: 1 fset3: Num of feature points: 52 10.000000[dpi] image. Size = (125,70) fset: Num of feature points: 1 fset3: Num of feature points: 41
Why is the number for initializing features the same?
fset contains the features for continuous tracking and fset3 the features for the recognition in the first place, right?