Object Detection Of Medical Image Based On Mathematical Morphology

Mathematical Morphology (MM) referred to as Morphology, is not only a theory, but also a powerful image analysis technique. The reason of applying to image analysis with Morphology is that Morphology’s purpose is to analyze the target’s shape and structure.Firstly, this paper describes the theory of Mathematical Morphology, and introduces a method of parameters estimation by using the fractional lower order moment and extreme value theory. On this basis, the paper studies kinds of medical image processing algorithms by applying Morphology.The main research work of this paper includes the following:(1)Introduce the background and significance of studying work, an overview of Mathematical Morphology, and Morphological image analysis, and describe the main studying work of this paper and chapters arrangement.(2)Briefly introduce the a-stable distribution and fractional lower order moment theory, and Use the two common parameters estimation algorithms (negative-order moment method and logarithmic order moment method) and a new parameters estimation algorithm (asymptotic extreme value method) to estimate the parameters α and γ of the two-dimensional wavelet coefficients of supersonic medical image. Asymptotic extreme value has the characteristics of rapid and real-time implement. Experimental result shows that the new method can also estimate the parameters α and γ of the two-dimensional wavelet coefficients which satisfy the symmetrical a-stable (SaS) distribution effectively.(3)Based on the theory of gray morphology, medical microscopic image filtering is to be achieved. First two kinds of two-dimensional structure elements are proposed, and remove the noise in gray image by using gray Morphology. The filtering effect has obvious differences with the difference of structure elements’type and size. Then uses these structure elements to carry on the image granularity detection, and draw the granularity distribution function to show the distribution status of image granules. The results show that two-dimensional granularity detection has the better effect than that in the one-dimensional detection and electing appropriate structure element plays an important role in image filtering and granularity detection.(4)Count the number of the objects (such as cell granules) interested to us in medical image. If all the granules in image are without any overlapping or crossing, every granule can be seen as a separate connected domain. Then mark every connected domain, and the statistic number of all the connected domains is the number of all the objects in image. In this case, any statistical error will not be happened. If some granules in image have slight overlapping with others, the statistic number can’t be obtained accurately through the above method. However, other image processing methods need to be used to operate on the overlapping granules, and separate these granules. Here using the theory of binary Morphology can eliminate overlapping details which influence statistical accuracy. Though the experiment, better effect can not only be got, but also improving the accuracy of granules statistic.(5)Finally, the research work of this paper is summarized and the further research work is pointed out.In short, Mathematical Morphology is a subject of image analysis, and it is widely used in image processing. Meanwhile, Morphology comes to show its superiority in image processing. This paper emphasizes on studying Morphological application to medical microscopic images.