The Theory Of Mathematical Morphology And It's Application In Image Processing

The Mathematical Morphology applies to every aspect related to Image Processing, such as target detection based on Hit-or-Miss, image segmentation based on the concept of watershed, skeleton abstraction and image coding compression based on erosion and open algorithm, image reconstruction based on Geodesic Distance, particle analysis based on morphological filter and so on. So the paper studied it deeply. The main tasks are as follows:(1)Researching into the theory of Mathematical Morphology, including some basic algorithms such as erosion, dilation, open, close and so on, and some advanced algorithms such as morphological gradient, Alternating Sequential Filter and so on. Discussing some image decomposition system such as The Discrete Size Transform, skeleton abstraction and image representation tool such as the pattern spectrum in detail.(2)Researching into the application of Mathematical Morphology in target detection. This part first introduces the method of Top-Hat and Bottom-Hat Transformation to enhance an image. Then researching into the principle of Morphological Reconstruction and discussing the problem in binary image reconstruction and grayscale image reconstruction respectively. Last, proving its practicability with feature detection in MR imaging and extraction of filarial worms from a microscopic image of blood stream.(3)Analyzing the application of Mathematical Morphology in image segmentation. For binary image segmentation, the paper discusses two methods: SKIZ and Watershed Transformation. SKIZ applies to binary non-overlapping particles image and Geodesic SKIZ applies to binary overlapping particles image. Watershed Transformation can get good segmentation result in both situation. For grayscale image segmentation, the paper adopts the method that combines internal marker and external marker to resist over-segmenting phenomenon successfully.(4)Analyzing the relationship between the Distance Transformation and Mathematical Morphology and bringing forward a novel adaptable mathematical morphology algorithm which can vary the size of structure element according to the characteristics of image dynamically. That's to say the algorithm can self-adapting adjust the size of structure element using the relationship between the Distance Transformation and Mathematical Morphology.A lot of experiments demonstrates on the one hand that the mathematical morphology has not only simple idea and algorithm realized easily but also good performance in image processing, and on the other hand the novel adaptable mathematical morphology algorithm presented in this paper has better adaptability than the traditional one.