Study On The Adaptive Filtering And Fast Algrithm Of Gray Scale Image Thresholding

Image segmentation has always been a hotspot in the field of computer science.With the rise of artificial intelligence,image segmentation has set off an upsurge of research.Image segmentation is an important foundation of image processing.Only when the imagesegmentation is well completed can the image will be better studied and applied.Threshold segmentation has become a research hotspot in image segmentation because of its simple and efficient characteristics.In the threshold segmentation method,the OTSU method,the minimum error method and the maximum entropy method are the three most classical methods.Among them,OTSU algorithm has been widely researched and applied because of its simple calculation,high real-time and no parameter.In general,the images obtained by various channels will contain a lot of noise due to various conditions and random disturbances,which will change the features of things in the original image.If this kind of image is analyzed directly,it will cause the deviation of image understanding,so the ability of image segmentation algorithm to suppress noise will be very important.The OTSU algorithm needs to search all the pixels when finding the optimal threshold,,and the amount of computation is very large.When the OTSU algorithm is extended to two dimensions,the complexity of computation is further improved,and the computation process is time-consuming,which is inconsistent with the real-time requirement in practical applications.Therefore,this dissertation studies the improvement of the segmentation efficiency and segmentation efficiency of OTSU algorithm.In this dissertation,for the most common salt and pepper noise in image,the median filtering algorithm is deeply studied.The improved median filtering algorithms(adaptive switch median filter algorithm and adaptive extrememedian filter algorithm)are applied to the OTSU algorithm,and two improved OTSU algorithms--OTSU algorithm based on adaptive median filter and OTSU algorithm based on adaptive extreme median filtering are obtained.The experimental results show that the algorithm can deal with the image polluted by salt and pepper noise better.The segmentation image is not only clear,but also can better retain the details of the image.Genetic algorithm is a search algorithm with high adaptability,fast operation speed and good global space search ability.It has a good effect when applied to the threshold of OTSU algorithm.But the traditional genetic algorithm will fall into the local optimal solution in varying degrees when the optimal threshold is obtained.Therefore,thisdissertation attempts to combine the improved genetic algorithm with the two-dimensional OTSU algorithm,and proposes the OTSU algorithm based on improved genetic algorithm.The functions in OTSU algorithm are used to guide optimal direction,and the excellent global optimization ability ofgenetic algorithm is used to find an optimal threshold in the image to distinguish the target and the background,and to complete the segmentation of the image.Experimental results show that the algorithm reduces the time required for segmentation and improves the segmentation efficiency,and the segmentation image can better retain the detail characteristics of the image.