Research On Cattle Follicle Detection Algorithms In Ultrasound Images Based On Combination Features And SVM

Ultrasound medical images are widely used in various stages of cattle reproduction and embryo production due to their advantages of fast,real-time,safety and low cost.In order to improve the reproductive capacity of cattle breeders,the application of ultrasound imaging technology to real-time monitoring of follicular changes in cattle plays a vital role,it is of great significance to increase the pregnancy rate.To achieve real-time monitoring of cattle follicles,the location of cattle follicles needs to be detected in the acquired ultrasound images.The main purpose of the research is to detect the area of bovine follicles in ultrasound images,so as to lay a foundation for further research and analysis such as real-time monitoring of cattle follicles.Aiming at the characteristics of cattle follicle ultrasound image,this article researchs the detection technology of follicle area in depth,firstly,relevant algorithms are studied to preprocess cattle follicle ultrasound image,namely denoising,based on the analysis and study of traditional BM3 D algorithm,PM model,Catte model and SRAD model,an improved anisotropic diffusion filtering algorithm is proposed,which achieves good denoising effect;After analyzing the defects of the traditional LBP algorithm,a new improved LBP algorithm-BALBP algorithm is proposed based on the research,at the same time,HOG and GLCM features of the image are extracted.In this article,PSO particle swarm optimization is used to optimize the parameters of SVM classifier,and AdaBoost decision tree classifier is used to optimize the weak classifier of SVM,so as to generate stronger classifiers with better ability,detection of bovine follicles based on combination features and SVM machine learning.Finally,through the analysis and comparison of the experimental results,it is verified that the method of improved combination features and optimized SVM machine learning classifier proposed in this paper has a good effect on the recognition rate and the efficiency of the algorithm implementation for detection of bovine follicles.