37
Distinctive Image Features from Scale-Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

Embed Size (px)

Citation preview

Page 1: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

Distinctive Image Features from Scale-Invariant Keypoints

Mohammad-Amin Ahantab

Technische Universität München, Germany

Page 2: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

2

Motivation

• Object recognition

• Motion tracking

• 3d scene recognition

• Stereo correspondence

• Panorama stitching…

Page 3: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

3

Which requirements should image features meet?

• Invariance to scale

• Invariance to rotation

• Invariance to affine transformation/view point

• Invariance to illumination

• High distinctiveness

Page 4: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

4

Scale Invariant Feature Transform (SIFT)

- Transforms image data to scale-invariant coordinates

- Creates large number of features

- A 500x500 pixel pictures results (roughly) in 2000 image features

Page 5: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

5

Scale Invariant Feature Transform (SIFT)

1. Scale-space extrema detection

2. Keypoint localization

3. Orientation assignment

4. Keypoint descriptor

Page 6: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

6

Scale-space

Page 7: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

7

How to blurr the image ?

Gaussian function:

Page 8: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

8

Next step : Calculate the Difference of Gaussian Images

𝐷 (𝑥 , 𝑦 ,𝜎 )=𝐿 (𝑥 , 𝑦 ,𝑘𝜎 )− 𝐿(𝑥 , 𝑦 ,𝜎 )

Page 9: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

9

Example

Page 10: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

10

Why is the Diffrence of Gaussian computed?

• Diffrence-of-Gaussian function is a good approximation to the scale-normalized Laplacian of Gaussian:

• Lindeberg (1994) : normalization of the laplacian with causes true scale invariance

• Mikolajczyk (2002) : extrema of produce the most stable image features

• Relation between DOG and can be explained by the heat diffusion equation:

Page 11: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

11

Why is the Diffrence of Gaussian computed?

• (finite differenceapproximation of )

• DoG function includes the factor required for scale invariance !

• Approximation much more efficient !

Page 12: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

12

Local Maximas/Minimas in DoG imagesLocating maxima and minimas by comparing the maked pixel with all its neighbour pixels in the current image, the scale above and the scale below it.

If the pixel is smaller or bigger than all of its 26 neighbours it will be selected.

Problem: only an approximation

Page 13: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

3D Ultrasound Mosaicing - Wachinger et al. 13

2.(Accurate) Keypoint localization

Bild durch Klicken auf Symbol hinzufügen

Page 14: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

. 14

Taylor expansion Bild durch Klicken auf Symbol hinzufügen

• The extremas (Samples) found in DoG images are not accurate

• They often times are located „between“ the pixels (subpixels)

• Accurate localization with the taylor expansion :

• The function‘s origin is at the sample point

Page 15: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

15

Taylor expansion

• Compute the location of the extremum : set the derivative of to zero

• If is smaller than 0.5 add this offset to location of the local sample point otherwise proceed with a diffrent sample point !

Page 16: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

16

Eliminate keypoints with low contrast !

Simply eliminate all extrema with the value less than 0.03 !

Page 17: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

17

Remove Edges!Bild durch Klicken auf Symbol hinzufügen

• Keypoints along edges are not stable !

• How edges are found: Large principal curvature across the edge, small one in perpendicular direction

• The principal curvature can be computed with the Hessian matrix of the DoG function because the eigenvalues of H are propotional to the curvatures!

Page 18: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

. 18

Calculation of principal curvatures

with

if r>10 eliminate key point

Page 19: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

19

3.Orientation assignment

• Compute magnitude and orientation of gradients around the keypoint on smoothed images L (x,y)

• Create A histogram with 36 bins (each 10 degrees ) and quantize with respect to magnitude

Page 20: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

20

• Take the orientation with the highest peak and assign it to keypoint !

• If there is a peak with at least 80% of the highest peak create a new keypoint with a diffrent orientation!

Page 21: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

21

Page 22: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

22

4.Keypoint descriptor

Page 23: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

23

4.Keypoint descriptor

• Image gradient magnitudes and orientations are sampled around keypoint in the region aound the keypoint

• A 16 x 16 sample array is used to compute a 4 x 4 descriptor• position and orientation of each sample is rotated relatively to key point

orientation (rotation invariance)

• In each field a histogram with 8 bins is created -> results in a 128 dim-vector

• Length of each arrow is computed with respect to the magnitude of the samples (Gaussian weight function )

Page 24: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

24

Illumination Invariance

• Changes in contrast : normalize the vector

• Changes in brightness : does not affect the gradients

• Non-linear illumination changes: limit the values in the vector to not larger than 0.2 and renormalize to unit length

Page 25: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

. 25

Key point matching

• matching keypoints by finding the nearest neighbour in a database of keypoints

• Nearest neighbour is the key point with the minimum Euclidean distance

• If the ratio between best and second best neighbour is bigger than 0.8 keypoints are not matched !

Page 26: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

26

Testig Sift for Image matching

Page 27: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

27

Page 28: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

28

Page 29: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

29

Page 30: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

30

Page 31: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

31

Page 32: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

32

Page 33: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

33

Page 34: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

34

Page 35: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

35

Page 36: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

36

Page 37: Distinctive Image Features from Scale- Invariant Keypoints Mohammad-Amin Ahantab Technische Universität München, Germany

37

References

• Lowe, D. “Distinctive image features from scale-invariant keypoints”

• http://www.aishack.in/tutorials/sift-scale-invariant-feature-transform-introduction/