Sparse Representation-based Analysis of Hyperspectral Remote Sensing Data

Transcription

1 Sparse Representation-based Analysis of Hyperspectral Remote Sensing Data Ribana Roscher Institute of Geodesy and Geoinformation Remote Sensing Group, University of Bonn 1

2 Remote Sensing Image Data Remote Sensing: Observing something from a distance without physical contact Infrared-Red-Green 2

3 Remote Sensing Image Data altigator.com, modified 3

4 reflectance Hyperspectral Image Data wavelength [μm] 4

5 Remote Sensing Tasks Unmixing Anomaly detection Classification Sparse representation-based analysis 5

6 Sparse Representation 6

7 Unmixing 7

8 Unmixing Task Processed satellite image Sub-pixel quantification? Pixel with class information (labeled) Pixel without class information (unlabeled) Endmember extraction Manually or Automatically Reconstruction by sparse representation Evaluation 8

9 Unmixing Task 1. task: Find suitable endmembers manually derived spectral library archetypal dictionary 2. task: Estimate fractions (activations) Sparse representation 9

10 Archetypal Analysis: SiVM Archetypal analysis finds the extreme points (archetypes) in feature space Efficient determination by Simplex Volume Maximization (SiVM) Assumption: Convex hull consists of points, which maximize the volume Christian Thurau, Kristian Kersting, Christian Bauckhage (2010): Yes We Can Simplex Volume Maximization for Descriptive Web-Scale Matrix Factorization 10

11 Archetypal Analysis: SiVM 1. Randomly choose (virtual) starting point 2. Choose sample which is farthest away 3. Set this sample as first archetype 4. Choose next sample which is farthest away from all previous archetypes Christian Thurau, Kristian Kersting, Christian Bauckhage (2010): Yes We Can Simplex Volume Maximization for Descriptive Web-Scale Matrix Factorization 11

12 Data Berlin-Urban-Gradient dataset

13 Data Study site: Southwest of Berlin Hyperspectral image Manually derived spectral library Reference land cover information Simulated EnMAP scene 13

14 Hyperspectral Data Airborne Sensor: HyMap 111 spectral bands Observed wavelength 450nm 2500nm Spatial resolution of 3.6m Visualized as RGB-image with the wavelengths R=640nm, G=540nm and B=450nm 14

15 Reference Information Reference information was manually obtained digital orthophotos cadastral data 4 land cover classes Impervious surface Vegetation Soil & Sand Water 15

16 Simulated EnMAP Data Simulated EnMAP scene of the same area Spatial resolution of 30m 1495 EnMAP pixels were obtained from the simulation tool, containing the fractions of the land cover classes ranging from 0 to 100% Task: Reconstruction of fractions of simulated EnMAP data 16

17 Archetypal Dictionary vs. Manually Derived Spectral Library Archetypal dictionaries were interpreted using reference data Archetypal dictionary Manually derived library Imp. Surface Vegetation Soil 2 4 Water High total amount of spectra in the manually derived spectral library 17

18 MAE [%] Evaluation Archetypal dictionary Manually derived library Imp. Surface Vegetation Soil Water Ø High number of elementary spectra in library results in a small reconstruction error All dictionaries achieve similar and satisfactory solutions Drees, L. and Roscher, R. (2017). Archetypal analysis for sparse representation-based hyperspectral sub-pixel quantification, ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., IV-1/W1,

19 Anomaly Detection 19

20 Anomaly Detection Task Dataset Detected symptoms Hyperspectral images (healthy and infected plants) 3D pointclouds No label information Dictionary elements extraction Sparse representation Standard dictionary Mixed topographic dictionary Evaluation 20

21 Biological Material Sugar beet plants (cultivar Pauletta; KWS GmbH, Einbeck, Germany) partially infected by the plant pathogen Cercospora beticola 21

22 Hyperspectral 3D Data Hyperspectral pushbroom sensor with 1600 pixel observing a spectral signature from 400nm to 1000nm Perceptron laser triangulation scanner Hyperspectral image Depth map Inclination map 22

23 Hyperspectral 3D Data Characteristic of hyperspectral data obtained from a healthy plant 23

24 Hyperspectral 3D Data Arrow direction: Difference of two wavelengths to reference signature (black dot) Arrow length: Value of difference Color (HSV): Angle and value of difference 24

25 Mixed Topographic Dictionaries Integration of prior knowledge into dictionary by means of inclination and depth groups Learned from healthy plant pixels Optimization with Group Orthogonal Matching Pursuit 25

26 Results Blue: specular reflections Green: leaf veins Orange: disease symptoms 26

27 Results Standard dictionary Mixed topographic dictionary Specular reflections Leaf veins Disease symptoms Roscher, R., Behmann, J., Mahlein, A.-K., & Plümer, L. (2016). On the Benefit of Topographic Dictionaries for Detecting Disease Symptoms on Hyperspectral 3D Plant Models, Workshop on Hyperspectral Image and Signal Processing. 27

28 Results Original image Reconstruction error index with standard dictionary Recontruction error index with mixed topographic dictionary Roscher, R., Behmann, J., Mahlein, A.-K., & Plümer, L. (2016). On the Benefit of Topographic Dictionaries for Detecting Disease Symptoms on Hyperspectral 3D Plant Models, Workshop on Hyperspectral Image and Signal Processing. 28

29 Classification with Self-taught Learning 29

30 Classification Task Processed satellite images Land use and land cover map Land use and land cover map Posterior probabilities Pixel with class information (labeled) Pixel without class information (unlabeled) Feature learning Classification Learning step Testing step Evaluation + Post-processing 30

31 Self-taught Learning training Dictionary contains unlabeled data Assumption: samples of the same class are reconstructed with a similar set of dictionary elements and similar weights Goal: new representation is highly discriminative 31

32 Feature/Representation Learning Learning a new data representation which is more suitable for a given task than the original data representation Powerful feature representation Discriminative Robust Lower complexity Easier to interpret 32

33 What is a Good Representation? Original representation Discriminative representation green class 1 feature 1 class 1 class 2 class 2 infrared feature 2 33

34 Self-taught Learning testing Dictionary is fixed Classifier is trained and tested with new representation 34

35 Self-taught Learning for Land Cover Classification Landsat 5 TM image near Novo Progresso (Brazil) Ca. 8000x8000 pixel 30x30m spatial resolution Area characterized by fire clearing Reference information: Forest, deforestation (fire clearing) and arable land Subarea: ~900km 2 35

36 Dictionary Choice ~1 Mio. image patches Archetypal analysis to decide on most important ones 36

37 Self-taught Learning for Land Cover Classification Satellite image Land cover Posterior probability: Maximum posterior arable probability (certainty) Roscher, R., Römer, C., Waske, B., Plümer, L. (2015). Landcover Classification with Self-taught Learning on Archetypal Dictionaries, IGARSS, Symposium Paper Prize Award 37

38 Summary Sparse representation is a versatile tool Exploitation of unlabeled samples for learning Self-taught learning Unsupervised learning (such as anomaly detection) More and more research goes into the direction of unsupervised pre-training in combination with supervised learning 38