A method of cattle follicle ultrasound images detection based on HOG + improved LBP + SVM

摘要

To improve the diagnostic accuracy of follicle ultrasound images detection, this paper proposed a method of cattle follicle ultrasound images detection based on HOG + Improved LBP + SVM. It calculated the Histogram of Oriented Gradient (HOG) feature for all cell in detection window, used improved Local Binary Pattern method to get gray feature, combined with the Support Vector Machine (SVM), it did the feature training and test experiment, last, the proposed method was compared with that single HOG feature detection single traditional LBP feature detection and HOG + traditional LBP feature detection. Experimental results showed, the proposed method can effectively describe and detect cattle follicle ultrasound images, and it has higher recognition accuracy.