BINARY G GENE NETIC P PROGRAMMI MING B BASED TEXTURE S SEGMENT NTATION
PROBLEM DEFINITION Texture segmentation with high accuracy is required for remote sensing and medical image analysis as it could be matter of life or death Current methods accuracy is Poor
RESEARCH QUESTIONS How is genetic programming performance compared to traditional methods of texture segmentation? How rich would genetic programming results will be to an ensemble of methods? Will utilizing multi genetic classifiers and a voting scheme will enhance genetic programming performance?
AGENDA
EDGE DETECTION Edge Detection Algorithm
LOCAL BINARY PATTERN: HISTOGRAM MATCHING Pixel Neighborhood Binary Representation
REGION GROWING
CO-OCCURRENCE MATRIX
LAW’S ENERGY MEASUREMENT • • L5 (Gaussian) gives a center-weighted S5 (LOG) detects spots • local average R5 (Gabor) detects ripples • E5 (Gradient) responds to row or col step edges
LAW’S ENERGY MEASUREMENT
Recommend
More recommend