Ce diaporama a bien été signalé.
Nous utilisons votre profil LinkedIn et vos données d’activité pour vous proposer des publicités personnalisées et pertinentes. Vous pouvez changer vos préférences de publicités à tout moment.

A rank based ensemble classifier for image classification

634 vues

Publié le

2013 8 TH I R ANIAN CONF E R ENC E ON MACHINE V I S ION AND IMAGE P ROC E S S ING (MVIP)

Publié dans : Sciences
  • Identifiez-vous pour voir les commentaires

A rank based ensemble classifier for image classification

  1. 1. A Rank based Ensemble Classifier for Image Classification using Color and Texture Features 2013 8 TH I R ANIAN CONF E R ENC E ON MACHINE V I S ION AND IMAGE P ROC E S S ING (MVI P ) FAT EMEH AHMADI 、MOHAMAD-HOS E YN S IGA R I 、MOHAMAD - E B R AHIM SHI R I
  2. 2. Outline Introduction Proposed Method ◦ Feature extraction ◦ Ensemble classifier ◦ Final decision maker Experimental Results Conclusion
  3. 3. Introduction
  4. 4. Image Classification contains two main steps: 1. Extraction of low-level features from input image. 2. Classification of input image based on the extracted features. Feature Extraction Classification A color image classification method using rank based ensemble classifier.
  5. 5.  Features:  Color: Color histograms are invariant to orientation and scale, and these properties makes it more powerful in image classification.  Texture: Texture is one of the most important characteristics of an image.  Classifiers:  Nearest Neighbor  Multi Layer Perceptron
  6. 6. a. Feature Extraction b. Ensemble Classifier c. Final Decision maker Proposed Method
  7. 7. Color Feature Texture Feature
  8. 8. A. Feature Extraction 1) Color Histogram: We extract color histogram in five color spaces: ◦RGB, HSV, CMY, YCbCr, 3D RGB quantize histogram in 10 bin for each color channel, therefore a feature vector of length 30 is acquired for each color space. Feature extraction -> Ensemble classifier -> Final decision maker
  9. 9. 2) Gabor Wavelet: Gabor wavelet operates like a local edge detector.  θ: determines orientation of the wavelet.  λ : specifies wavelength of cosine signal.  ψ: is phase of the cosine signal.  σ: denotes radius of the Gaussian function.  ϒ: specifies aspect ratio of the Gaussian function. Feature extraction -> Ensemble classifier -> Final decision maker
  10. 10. In the proposed system: ◦ rotation angles : {0, π/4, π /2, 3 π /4} ◦ Wavelengths: {2,2 2 ,4} There are 12 different Gabor filters. After convolving the image by all Gabor filters, 12 2D coefficient matrixes are obtained, which are denoted by Ci while i {1,…,12}. a) The First Feature set: ◦ Histogram of AM and counts the dominant edges in different width and orientation. b) The Second Feature Set: ◦ We compute mean and variance of coefficient matrixes. Therefor, length of the second texture feature set is 24 for each image. Feature extraction -> Ensemble classifier -> Final decision maker
  11. 11. B. Ensemble Classifier Do not learn a single classifier but learn a set of classifiers combine the predictions of multiple classifiers. (https://www.ke.tu-darmstadt.de/lehre/archiv/ws0405/mldm/ensembles.pdf) Supplement We use two classifiers as simple classifier in ensemble: (1) Nearest Neighbor (NN) ◦ Class labels of these 3 nearest neighbors are listed as output in an ordered list (2) Multi Layer Perceptron (MLP) ◦ Output of MLP is an ordered list of 3 classes that have higher values in the corresponding neurons in output layer Feature extraction -> Ensemble classifier -> Final decision maker
  12. 12. Proposed System: In the Color(5): RGB, HSV, YCbCr ,CMY, 3D RGB  Texture(2): Dominant edges and statistical moments of Gabor coefficients.  Classifier(2):NN, MLP (5+2)*2 = 14 Therefore, our ensemble classifier contains 14 simple decision makers Feature extraction -> Ensemble classifier -> Final decision maker
  13. 13. C. Final Decision Maker To combine outputs and make the final decision in an ensemble classifier.  Simple majority vote ◦ all simple decision makers have equal importance in the ensemble.  Weighted majority vote ◦ the importance of each simple classifier is different and votes of each classifier is weighted by a coefficient in range of (0,1) Feature extraction -> Ensemble classifier -> Final decision maker
  14. 14. Experimental Results
  15. 15. Experimental Results Corel dataset:  1000 images  10 Classes.(each class contains 100 images) In each test iteration, 100 images of 1000 images are used as test data and the remainders are used as training data. Therefore, test iterations are repeated for 10 times.
  16. 16. A. Experiments on Simple Decision Makers
  17. 17. B. Experiments on The Ensemble Classifier using SimpleMajority Vote We compare two different conditions for majority vote: (1) using only one label as output of each simple decision maker (2) using 3 labels as output of each simple decision maker.
  18. 18. C. Experiments on The Ensemble Classifier using WieghtedMajority Vote
  19. 19. Conslusion & Feature work
  20. 20. Rank based ensemble classification of extracted feature sets work very good for color image classification. For improvement of the proposed system, we suggest to use other features like shape base features and other classifiers like decision tree and Support Vector Machine (SVM). Additionally, proposing an adaptive method for weighting of ordered list of labels may lead to achieve a more robust and efficient system for image classification.
  21. 21. End Thank you

×