A Research On Image Processing Method Based On Slope Difference Distribution And Morphological Operation

With the rapid development of computer technology,many industrial production,automation equipment,medical equipment and so on all need mature image processing technology,and the research on image processing technology has become one of the hot spots in today's academic research.Various image processing methods: traditional algorithms based on image features or deep learning algorithms have been widely used in academic and market applications,but they also have their corresponding defects.In recent years,researchers have been constantly exploring an image processing method with high automation level,accurate segmentation and strong robustness.The research content of this paper based on these three characteristics.Through the application of image processing in medical images,a threshold selection method based on slope difference distribution is adopted.On cell adhesion in image is proposed based on morphological area constraints limit corrosion treatment method,in view of the heart in the left ventricle MRI image of left and right ventricular adhesion,such problems as lack of myocardium and the ventricular hypertrophy was proposed based on Hough transform method to extract the left ventricle,and through the contrast experiment with a variety of image processing methods,the effectiveness of the proposed method was verified.The main research contents of this paper include as follow:(1)For cell images,compared the slope difference distribution threshold selection method with other eight threshold selection methods to verify the effectiveness of the slope difference threshold method.(2)Through the basic operation of morphology,a morphological area-constrained ultimate erosion method for the region constraint of adhesion cells are proposed in this paper,which can effectively segment the adhesion cells and has higher quantitative results than other segmentation algorithms.(3)Aiming at the phenomenon of uneven illumination and gray intensity distribution in the image,we utilize Gabor filtering to perform gradient filtering of the image,and a more robust cell boundary can be segmented by SDD method,and in this paper,the sensitivity analysis of Gabor filter parameters is carried out by means of control variables in order to get best Gabor filter parameters.(4)Based on the left ventricle MRI image,a fully automatic segmentation algorithm based on Hough transform circle detection(CHT)and double threshold segmentation of the myocardium are proposed in this paper.By restricting the region of the circle detection to the segmented result,partition accuracy can achieve above 94%.