Image Segmentation Under Complex Background And Its Application

Image segmentation plays an important role in the field of computer vision and image processing,which is the basis of subsequent image application.For the complex background image gained from the drone cruise or patrol,with the feature of image random shooting position and uneven light,it is necessary to improve the existing image segmentation algorithm to get the ideal segmentation effect for providing a guarantee for the follow-up target monitoring or tracking.As drone and other embedded devices demand that the image processing algorithm must to be written in a more basic way,the establishment of an image segmentation algorithm model library to provide support has an important practical significance.The image segmentation method based on the active contour model is a hot topic in the field of image segmentation in recent years,and it shows a better segmentation characteristic compared with the traditional contour extraction algorithm.This thesis study the algorithm of segmentation method based on active contour model,and propose an improved Snake model method based on watershed algorithm to provide a better segmentation solution for computer vision and image processing.The main research and achievement present as follows:(1)The simulation experiments of the classical image segmentation algorithm are designed,among them,the segmentation based on threshold and the segmentation based on region are studied,and the underlying algorithms are implemented in C ++ language,verifying the correctness of these algorithms.(2)The image segmentation methods of geometric active contour model and parametric active contour model are studied in this thesis.This thesis focuses on the segmentation method of C-V model and classical Snake model,and analyzes the advantages and disadvantages of algorithm computing and model constructing.On the basis of this study,the improved GVF-Snake model is studied and the algorithm is designed and implemented.Eventually suitable simulation experiments have been done to show the feature of the active contour model.(3)Aiming at the image segmentation under complex background,an improved Snake model combining with watershed algorithm is proposed.In this thesis,the Snake model is used to provide the initial contours of the real contours by using the watershed algorithm,and the simulation results are used to verify the complexity of the algorithm and achieve better segmentation results.(4)An image segmentation algorithm library,based on Qt frame,which is designed to deal with complex background,is developed in this thesis.Segmentation algorithms in this library are written by C + + language and appropriate design pattern for the software structure is designed to ensure the system stability.