Classification Using Extended Morphological Attribute Profiles Based On Different Feature Extraction Techniques  Stiijn Pe...
Overview <ul><li>Background </li></ul><ul><li>Morphological Attribute Profiles </li></ul><ul><li>Classification of Hypersp...
Background <ul><li>Morphological profiles (MP) and Morphological attribute profiles (MAP) have been successfully used to f...
Background <ul><li>Traditionally, MPs and MAPs are built using the feature extraction based on principal component analysi...
Morphological Profile (MP)  and  Morphological Attribute Profile (MAP)
Morphological Profiles When dealing with real images it is difficult to identify a single filter parameter suitable to han...
Morphological Profiles Closing Profile Opening Profile Square SE Sizes: 7, 13, 19, 25 Morphological Profiles (MPs) are com...
X 1 X 1 X 1 X 1 MP X 1 Hyperspectral Image X MP MP X 1 Morphological profile  1 Morphological profile  n Feature  Reductio...
Attribute Profiles Thickening Profile Thinning Profile Square SE  (MP) Sizes: 7, 13, 19 Area Attribute λ : 45, 169, 361 Cr...
Selection of thresholds for constructing MAP Master Thesis: Mattia Pedergnana (University of Iceland, Iceland and Universi...
Results
Data used ROSIS University of Pavia
Data used AVIRIS Indian Pine
Data used <ul><li>ROSIS University of Pavia </li></ul>AVIRIS Indian Pines Training Test Trees 524 2912 Asphalt 548 6304 Bi...
Results: University of Pavia data PCA SVM OA  92.01% AA 92.17% k 0.8957 Random Forest OA  91.31% AA 87.96% k 0.8894
Results: University of Pavia data Kernel PCA RF OA  92.2% AA 95.02% k 0.8993 SVM OA  92.31% AA 93.96% k 0.9002
Results: University of Pavia data DAFE RF OA  96.25% AA 96.28% k 0.951 SVM OA  92.69% AA 93.27% k 0.9119
Results: University of Pavia data DBFE RF OA  95.09% AA 95.32% k 0.9386 SVM OA  93.45% AA 94.16% k 0.9145
Summary of Results University of Pavia Extended Attribute Profile using Standard deviation RF SVM PCA KPCA DAFE DBFE PCA K...
Extended Morphological Profiles Params: Initial Size: 1 Step: 2 Number Of Opening/Closing: 3 PCA RF DAFE RF DBFE RF KPCA R...
Results: Indian Pine PCA RF OA  92.20% AA 95.32% k 0. 9100 SVM OA  88.94% AA 92.96% k   0.8738
Results: Indian Pine Kernel PCA RF OA  92.87% AA 96.01% k 0.9183 SVM OA  88.93 % AA 93.36 % k   0.8737
Results: Indian Pine DAFE RF OA  77.21   % AA 87.62   % k 0.7427 SVM OA  68.70% AA 71.31% k 0.6464
Results: Indian Pine DBFE RF OA  81.28% AA 87.78% k 0.7866 SVM OA  73.13% AA 79.81% k   0.6954
Summary of Results Indian Pine Extended Attribute Profile using Standard deviation RF SVM PCA KPCA DAFE DBFE PCA KPCA DAFE...
Extended Morphological Profiles Params: Initial Size: 1 Step: 2 Number Of Opening/Closing: 3 PCA RF DAFE RF DBFE RF KPCA R...
Conclusion <ul><li>The results of classifying hyperspectral data using morphological attribute filters with various featur...
Conclusion <ul><li>However, non-linear techniques such as kernel PCA preserve the information of the independent clusters ...
Experimental Results <ul><li>Pavia Dataset – Second Approach – HML-DBFE-RF </li></ul>Overall Accuracy: 98.3%  Average Accu...
Experimental Results <ul><li>Pavia Dataset – Second Approach – HML-KPCA-RF </li></ul>Overall Accuracy: 90%  Average Accura...
Experimental Results <ul><li>Pavia Dataset – Second Approach – HML-KPCA-SVM </li></ul>Overall Accuracy: 93.8%  Average Acc...
Experimental Results <ul><li>Indian Pine– Second Approach – HML-KPCA-RF </li></ul>Overall Accuracy: 93.8%  Average Accurac...
<ul><li>Thank You for your attention </li></ul>Software in Matlab can be freely obtained by sending us an email. Prashanth...
Upcoming SlideShare
Loading in …5
×

IGARSS_2011_MARPU_3.ppt

1,109 views
1,017 views

Published on

0 Comments
1 Like
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total views
1,109
On SlideShare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
43
Comments
0
Likes
1
Embeds 0
No embeds

No notes for slide

IGARSS_2011_MARPU_3.ppt

  1. 1. Classification Using Extended Morphological Attribute Profiles Based On Different Feature Extraction Techniques Stiijn Peeters University of Antwerp, Antwerp, Belgium Dr. Prashanth Marpu Mattia Pedergnana Prof. Jon Atli Benediktsson University of Iceland, Reykjavik, Iceland Dr. Mauro Dalla Mura Prof. Lorenzo Bruzzone University of Trento, Trento, Italy
  2. 2. Overview <ul><li>Background </li></ul><ul><li>Morphological Attribute Profiles </li></ul><ul><li>Classification of Hyperspectral data </li></ul><ul><li>Ongoing work </li></ul><ul><li>Conclusions </li></ul>
  3. 3. Background <ul><li>Morphological profiles (MP) and Morphological attribute profiles (MAP) have been successfully used to fuse spectral and spatial information for the classification of remote sensing data. </li></ul><ul><li>J. A. Benediktsson, J. A. Palmason, and J. R. Sveinsson, “Classification of Hyperspectral Data From Urban Areas Based on Extended Morphological Profiles,” IEEE Trans. Geosci. Remote Sens. , vol. 43, no. 3, pp. 480–490, Mar. 2005 </li></ul><ul><li>M. Dalla Mura, J. A. Benediktsson, B. Waske, and L. Bruzzone, ”Morphological Attribute Profiles for the Analysis of Very High Resolution Images,” IEEE Trans. Geosci. Remote Sens. , vol. 48, no. 10, pp. 3747–3762, Oct. 2010 </li></ul>
  4. 4. Background <ul><li>Traditionally, MPs and MAPs are built using the feature extraction based on principal component analysis (PCA). </li></ul><ul><li>Moreover, the selection of filter parameters is traditionally done manually. </li></ul><ul><li>In this study, </li></ul><ul><li>We analyse the classification results by using various feature extraction techniques (PCA, Kernel PCA, DAFE, DBFE). </li></ul><ul><li>We use a simple method to build the MAPs based on standard deviation attribute automatically. </li></ul>
  5. 5. Morphological Profile (MP) and Morphological Attribute Profile (MAP)
  6. 6. Morphological Profiles When dealing with real images it is difficult to identify a single filter parameter suitable to handle all the objects in the image. Perform a multilevel analysis by using several values for the filter parameters. Build a stack of images with different degrees of filtering. Morphological Profile (MP) M. Pesaresi and J. A. Benediktsson, “A new approach for the morphological segmentation of high-resolution satellite imagery,&quot; IEEE Transactions on Geoscience and Remote Sensing , vol. 39, no. 2, pp. 309-320, 2001.
  7. 7. Morphological Profiles Closing Profile Opening Profile Square SE Sizes: 7, 13, 19, 25 Morphological Profiles (MPs) are composed by a sequence of opening and closing with SE of increasing size. MP
  8. 8. X 1 X 1 X 1 X 1 MP X 1 Hyperspectral Image X MP MP X 1 Morphological profile 1 Morphological profile n Feature Reduction F 1 F 2 F n X 1 X 1 Extended Morphological Profile X 1 X 1 X 1 X 1 Extended Morphological Profile
  9. 9. Attribute Profiles Thickening Profile Thinning Profile Square SE (MP) Sizes: 7, 13, 19 Area Attribute λ : 45, 169, 361 Criterion : Area > λ Moment of Inertia Attribute λ : 0.2, 0.1, 0.3 Criterion : Inertia > λ STD Attribute λ : 10, 20, 30 Criterion : STD > λ
  10. 10. Selection of thresholds for constructing MAP Master Thesis: Mattia Pedergnana (University of Iceland, Iceland and University of Trento, Italy) Optimal Automatic Construction of Morphological Profiles In this study, we only use the attribute profile generated using the standard deviation attribute. The thresholds to build the profile are estimated for every feature separately from the range of standard deviation values of the training samples of all the classes. So, different threshold values are used for diferent profiles. A more general approach to use a big range of attributes has been recently proposed. An entire profile using a wide range of attributes and wide range of thresholds is built and a newly proposed hybrid genetic algorithm is used for feature selection.
  11. 11. Results
  12. 12. Data used ROSIS University of Pavia
  13. 13. Data used AVIRIS Indian Pine
  14. 14. Data used <ul><li>ROSIS University of Pavia </li></ul>AVIRIS Indian Pines Training Test Trees 524 2912 Asphalt 548 6304 Bitumen 375 981 Gravel 392 1815 Metal sheets 265 1113 Shadow 231 795 Bricks 514 3364 Meadows 540 18146 Bare soil 532 4572 Training Test Corn-notill 50 1384 Corn-mintill 50 784 Corn 50 184 Grass-pasture 50 447 Grass-trees 50 697 Hay-windrowed 50 439 Soybean-notill 50 918 Soybean-mintill 50 2418 Soybean-clean 50 564 Wheat 50 162 Woods 50 1244 Bld-Grass-Trees-D 50 330 Stone-Steel-Towers 50 45 Alfalfa 15 39 Grass-pasture-mowed 15 11 Oats 15 5
  15. 15. Results: University of Pavia data PCA SVM OA 92.01% AA 92.17% k 0.8957 Random Forest OA 91.31% AA 87.96% k 0.8894
  16. 16. Results: University of Pavia data Kernel PCA RF OA 92.2% AA 95.02% k 0.8993 SVM OA 92.31% AA 93.96% k 0.9002
  17. 17. Results: University of Pavia data DAFE RF OA 96.25% AA 96.28% k 0.951 SVM OA 92.69% AA 93.27% k 0.9119
  18. 18. Results: University of Pavia data DBFE RF OA 95.09% AA 95.32% k 0.9386 SVM OA 93.45% AA 94.16% k 0.9145
  19. 19. Summary of Results University of Pavia Extended Attribute Profile using Standard deviation RF SVM PCA KPCA DAFE DBFE PCA KPCA DAFE DBFE OA% 91.31 92.2 96.28 95.32 92.01 92.31 93.27 93.45 AA% 87.96 95.02 96.25 95.09 92.17 93.96 92.69 94.16 k 0.886 0.899 0.951 0.939 0.896 0.90 0.912 0.914
  20. 20. Extended Morphological Profiles Params: Initial Size: 1 Step: 2 Number Of Opening/Closing: 3 PCA RF DAFE RF DBFE RF KPCA RF PCA SVM DAFE SVM DBFE SVM KPCA SVM AA (%) 90.89 95.12 96.75 93.87 91.37 92.32 94.92 94.74 OA (%) 84.75 94.88 96.107 87.89 86.92 88.93 90.00 91.47 k 0.8048 0.9324 0.9487 0.844 0.8326 0.8579 0.8720 0.8898
  21. 21. Results: Indian Pine PCA RF OA 92.20% AA 95.32% k 0. 9100 SVM OA 88.94% AA 92.96% k 0.8738
  22. 22. Results: Indian Pine Kernel PCA RF OA 92.87% AA 96.01% k 0.9183 SVM OA 88.93 % AA 93.36 % k 0.8737
  23. 23. Results: Indian Pine DAFE RF OA 77.21 % AA 87.62 % k 0.7427 SVM OA 68.70% AA 71.31% k 0.6464
  24. 24. Results: Indian Pine DBFE RF OA 81.28% AA 87.78% k 0.7866 SVM OA 73.13% AA 79.81% k 0.6954
  25. 25. Summary of Results Indian Pine Extended Attribute Profile using Standard deviation RF SVM PCA KPCA DAFE DBFE PCA KPCA DAFE DBFE OA% 92.20 92.87 77.21 81.28 88.94 88.93 68.70 73.13 AA% 95.32 96.01 87.62 87.78 92.96 93.36 71.31 79.81 k 0.910 0.918 0.743 0.787 0.874 0.8737 0.646 0.695
  26. 26. Extended Morphological Profiles Params: Initial Size: 1 Step: 2 Number Of Opening/Closing: 3 PCA RF DAFE RF DBFE RF KPCA RF PCA SVM DAFE SVM DBFE SVM KPCA SVM AA (%) 93.54 87.67 86.69 95.66 92.90 69.84 78.93 92.702 OA (%) 87.75 79.08 78.98 92.01 87.15 66.61 71.36 86.67 k 0.8607 0.7627 0.7615 0.9085 0.8533 0.6231 0.6763 0.8483
  27. 27. Conclusion <ul><li>The results of classifying hyperspectral data using morphological attribute filters with various feature extraction techniques has been studied. </li></ul><ul><li>Better results are obtained with attribute profiles compared to morphological profiles. </li></ul><ul><li>Supervised feature reduction techniques are constrained by the number of available samples and hence do not provide consistent results. </li></ul><ul><li>Linear unsupervised feature reduction techniques such as PCA may not be useful as the features are not always able to distinguish between classes effectively. This is observed in the experiments. </li></ul>
  28. 28. Conclusion <ul><li>However, non-linear techniques such as kernel PCA preserve the information of the independent clusters and hence can be useful in distinguishing between classes. Experiments suggest that EAP with KPCA produces consistent and high quality classification results. </li></ul><ul><li>We are currently investigating the results with a wide range of feature extraction techniques. </li></ul><ul><li>We are also working on methods to automatically identify the thresholds to build profiles. </li></ul>
  29. 29. Experimental Results <ul><li>Pavia Dataset – Second Approach – HML-DBFE-RF </li></ul>Overall Accuracy: 98.3% Average Accuracy: 98.6%
  30. 30. Experimental Results <ul><li>Pavia Dataset – Second Approach – HML-KPCA-RF </li></ul>Overall Accuracy: 90% Average Accuracy: 97 %
  31. 31. Experimental Results <ul><li>Pavia Dataset – Second Approach – HML-KPCA-SVM </li></ul>Overall Accuracy: 93.8% Average Accuracy: 96.9 %
  32. 32. Experimental Results <ul><li>Indian Pine– Second Approach – HML-KPCA-RF </li></ul>Overall Accuracy: 93.8% Average Accuracy: 96 %
  33. 33. <ul><li>Thank You for your attention </li></ul>Software in Matlab can be freely obtained by sending us an email. Prashanth Marpu [email_address] Mattia Pedergnana [email_address] Mauro Dalla Mura [email_address]

×