Research And Application Of Otsu Algorithm Based On Median In Image Processing

Image processing techniques applied in many fields such as military, medicine,biology, especially in the military field application is particularly important.In recentyears,many countries have been coveted the rich ocean areas and resources of China,especially Japan, its ships have never stopped harassing the waters in our country.Therefore, to strengthen both the capacity of monitoring China’s territorial sea andreal-time detecting the invading ships remains vital.To meet the requirements of moderninformation war, it is very necessary to monitor the territorial waters so that we can getsuspected target image information. In particular, automatic interpretation andidentification through computer is very important, it can effectively improve theaccuracy of the military intelligence and real-time performance, reduce the chance ofwrong identification of target, which can give a strong help for our military strength.With the application of the technology of image segmentation and feature matching incomputer vision, It is now possible to gain the information that we need fromreconnaissance image, at the same time,the technology also promoted the level ofChina’s military reconnaissance technic.The detection of target, the discrimination of characteristics, and the technology ofautomatic identification in military image have always been popular issues, but alsoissues that difficult to be solved nowadays in the modern research of militaryinformation technology. However, these technologies have a wide application fieldmainly reflected in military intelligence gathering, enemy monitoring and automaticguided in weapons. If these technologies can be full used, China’s military power willbe greatly enhanced. Among these technologies, the processing and recognitiontechnology of ship’s target image that showed in satellite images, which based on a highrecognition rate (including space recognition rate, time recognition rate and spectrumidentification ratio. Here mainly refers to space recognition rate.), holds an important application prospect in the preparation of military battle, especially in the sovereigntyover DiaoYu Island.Threshold segmentation is always a basic problem in the field of image processing,even in the era of high development of science and technology today, it still remains asa hot research direction. Its first aim is to separate the goals that we need from thecomplex background pictures. There are many kinds of algorithms, of which two aremost commonly used, namely, global threshold segmentation algorithm and localthreshold segmentation algorithm. Of the two algorithms, the global thresholdsegmentation algorithm is comparable simple and easy to implement, and real-timeperformance is high and the effect is ideal. But because of this algorithm is based on thewhole picture, disadvantage is equally obvious, that is: if there are partial changes in theimages then this algorithm is unable to achieve a rapid response. In contrast, the basicidea of the local threshold segmentation algorithm is making local processing. So whenpartial changes happen in the images, it could response well and with a good robustness.Still but, this algorithm has a trivial operation, it not only needs man set parameters in itbut also takes a long time in processing. This essay gives a following researchaccording to the background images and the characteristics of the targets that we need:Firstly, eliminate the noise of the original image, and increase the contrast. Thenpreprocess the processed images, and we can get the result of the first processing;secondly, analyze different kinds of the classical threshold segmentation algorithm andfind out the advantages and disadvantages of each algorithm; thirdly, after the study ofOtsu threshold segmentation method, we can conclude that: even though the classicsegmentation algorithm effects well, its complicated operation, time consuming alsomakes its obvious shortcomings; fourthly, we can describe characteristics of Otsusegmentation method, which based on median, through the analysis of all thedisadvantages that exposed by binary image histograms. This method could get anaccurate result and a short operation time as its advantages, but a partial missing targetpixels as its drawback.In conclusion, this paper puts forward the emphasize-on-the-bottom method, thismethod makes real target and background have a biggest difference, which helps us that when extracting target in the background image, we can gain the separated structureeasier and more accurate. Therefore, we can apply the proposed algorithm in themilitary field, and give a brief introduction of the involved technology.We have had afully experiment on the Otsu algorithm based on median. It can be found from theexperimental results that the improved algorithm in this paper, for most important, ismore robust, and judged by the extracted target images, its segmentation effect isparticularly prominent. If it could be developed and applied in the military field, it willplay a vital role in the field of military target tracking and separation, and we will havea further development in image processing field.