Using Association Rules as Texture features Authors: J.A. Rushing, H.S. Ranganath, T.H. Hinke, and S.J. Graves Source: IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 23, No. 8, pp. 845-858 Speaker: Tzu-Chuen Lu
Outline Introduction Image classification and segmentation Association rules for image data Association rules for texture classification Experimentations Conclusions
Image Classification ?
Image Segmentation
Image features
Image Classification
Association Rules
Association Rules
Association rules for image data X = 0 X = 1 X = 2 X = 3 X = 4 Y = 0 Y = 1 Y = 2 Y = 3 Y = 4 N*N image
Association rules for image data 3*3 pixels Root pixel 1 2 3 4 N*N image
Association rules for image data 1-Item: (X, Y, I)
Association rules for image data {(0, 0, 0), (1, 0, 2)} {(0, 0, 2), (1, 1, 2)}
Association rules for image data
Association rules for image data Sup ({(0, 0, 0)}) = 3, Sup ({(1, 0, 2)}) = 4
Association rules for image data Min confidence = 1
Texture 1 Texture 2 Texture 3 Texture 4
Texture 1 Texture 2 Texture 3 Texture 4
Suite 1: man made textures Suite 2: natural textures Suite 3 : suite 1 + suite 2 Train samples: 32 Test samples: 32
Association rules for texture classification
Association rules for texture classification
Image Segmentation
(a) Association Rules (b) Gabor filter (c) GLCM Image Segmentation (a) Association Rules (b) Gabor filter (c) GLCM
Conclusions New texture features based on association rules Classification and segmentation Time complexity