Research On Intelligent Image Segmentation Algorithm Based On Grey Wolf Optimization

As one of the main ways for information expression,transformation and communication,the image plays an important role in our human life and industry.For example,the surveillance video provides effective guarantee of traffic safety and crime fighting,the medical images are the solid basis of disease diagnosis and treatment.Image information processing and analysis are the founding of the mentioned guarantee and basis,while the image segmentation taken as the first step of image processing appears to be particularly important.The effect of image segmentation will directly affect the acquisition of image information,therefore the research of image segmentation has the great theoretical significance and wide application prospect.This thesis models the process of image segmentation as a parameters optimization problem and uses the modified Gray Wolf Optimizer to optimize the objective function.The specific work is as follows:First,proposing to apply the weight-based Improved Discrete Gray wolf Optimizer(IDGWO)to image segmentation.The thesis improves the Grey Wolf Optimizer(GWO)by involving in the concept of weight for updating the search location and combines the improved algorithm with Kapur entropy segmentation method for image segmentation.The simulated experiments prove that this optimizer has advantages in optimizing the image segmentation.Second,raising an improved Gray Wolf Algorithm based on Artificial Bee Colony(Fused Gray Wolf Optimizer,FGWO).The thesis makes further improvements and fusion to Gray Wolf Optimizer by applying the concept of local search and greedy selection of optimal solution in Artificial Bee Colony(ABC)algorithm to the same sections of GWO.The fused algorithm has been tested on benchmark functions and the results of it show that the improved version has distinct effects when optimize complex multi-peaks function with high dimensions.Third,we apply the Fused Gray Wolf Optimizer(FGWO)to the image segmentation which focuses on medical images.The combination of FGWO algorithm and Kapur entropy segmentation method is used for medical image segmentation and plenty of simulated experiments are tested on magnetic resonance(MR)images of the human brain.Experimental results indicate that the application of FGWO to medical image segmentation can produce good segmentation.This integrated approach possesses high operability and it is a research field with important value.