【作者】 李盛阳；

【导师】 孙忠林；

【作者基本信息】 山东科技大学 ， 计算机应用技术， 2003， 硕士

【摘要】 本文讨论了数字图象处理技术在肺部细胞图象上的应用。首先进行细胞图象增强处理。细胞图象增强主要目的是改善细胞图象的质量，突出细胞图象的整体或局部特征，提高细胞的视觉效果和识别特征。细胞图象增强主要采用灰度变换、直方图修正、图象平滑、图象锐化等方法。灰度变换调整改变细胞图象的灰度，突出了感兴趣的区域。通过直方图修正得到均衡化或规定化等所需的不同的处理效果。采用有效的图象平滑方法对细胞图象进行平滑降噪处理，消除图象数字化和传输时所混入的噪声，提高了图象的视觉效果。利用图象锐化处理突出细胞的边缘信息，加强细胞的轮廓特征，以便于一些基本特征的提取，并比较各种图象锐化方法处理细胞图象时的结果。 其次，对图象的分割方法包括细胞图象的分割方法进行了较详细的综述，指出现有方法的实现原理与特点。本文重点研究了数学形态学在二值细胞图象和灰度细胞图象分割处理中的应用。对二值图象数学形态学和灰度图象数学形态学的基本理论进行了较详细的介绍，利用二值图象数学形态学的边缘形态梯度检测出二值细胞图象的边缘，并对该方法进行了一定的研究；对于灰度图象，利用形态梯度、Top-Hat变换等进行细胞的边缘检测。对灰度图象数学形态学的边缘检测方法进行了改进，应用多结构多尺度结构算子的方法，达到既能检测细胞边缘又能有效降低噪声的目的。针对细胞之间存在的粘连和重叠现象，对基于数学形态学中的开闭运算和流域分割的方法进行了研究以便于进行细胞之间的分离。 论文最后讨论了细胞图象的分析。利用局部邻域或图象标号等方法进行细胞计数。通过形态特征的提取和选择，利用最小距离法和形态法对肺部细胞进行了初步的识别分类。更多还原

【Abstract】 This paper mainly discussed the digital image processing techniques in application of lung cell image. Firstly, to process the cell image so as to enhance the quality of the image, the aim of enhancing the cell image is to give prominence to the whole and portion feature of the cell image and improve the effect of vision and the feature of recognition. The method of enhancing the cell image include gray-scale transformation, histogram modification, image smoothing and image sharpening etc. gray-scale transformation adjust the gray-scale of the cell’s image and give prominence to the interested area. By use of the method of histogram modification can get the results of equipoise and prescript. The method of image smoothing to process the cell image can eliminate the noises that caused by digital process and transmission and improve the effect of vision. The method of image sharpening give prominence to the edge information of the cell’s image and strengthen the feature of contour so as to the extraction of basic feature, at last by comparison the results that were processed by some kinds of method of image sharpening can get which method is suit for this type of cell’s image.Secondly, there is a survey about image segmentation method including the cell’s image segmentation method, the survey indicate the principle and the characteristic of the segmentation methods. In this paper, the emphasis is the segmentation application of mathematical morphology to binary cell’s image and gray-scale cell’s image. There is a detailed presentation about the basic theory of binary image mathematical morphology and gray-scale image mathematical morphology. Using the edge modal detection method of binary image mathematical morphology can detect the edge of binary cell’s image. In this paper, there is a research about the edge modal detection method of binary image mathematical morphology. As for gray-scale image, using the method of modal grads and Top-Hat transformation can detect the edge of cells. The edge detection method of gray-scale image mathematical morphology was be improvedbecause of the disadvantage. Using the multi-structure and multi-scale operator to process the cell image can not only detect the edge of cell but also effectively reduce the noise. Aim at the conglutination and overlap in cells, researched the method of the opening and closing operation and valley segmentation based on mathematical morphology.Finally, discussed the cell image’s analysis. By using the local neighborhood algorithm or the image label algorithm can take count of the number of cells. By extraction and selection of the cell shape feature, using the least distance method and shape method primarily identify and classify the lung cell.更多还原