The Segmentation Of Cell Images Based On The Morphological Watershed Algorithm

The paper focus on colored and grayscale image, in view of conducts the research based on the morphology watershed algorithm's cell image division.First, the background and significance of the study on image segmentation are explained. The status of the cells image analysis system at home and abroad is introduced. The direction of developments of cells image segmentation technology in this field discussed. The concept of image segmentation and the characteristics of cells image as well as the existing segmentation methods of analysis are reviewed.Secondly, I focus on the cells image segmentation method using the watershed algorithm based on the morphology. I present the basic theory of morphology; present the principle and the application of morphological reconstruction; present the essential principle of morphological watershed algorithm and the common method to resolve the problems of overlap in detail.Due to the number of advantages that it possesses, the watershed transform has been widely used in many fields of image processing. It can be parallelized performance, and it produces a complete division of the image in separated regions even if the contrast is poor, thus avoiding the need for any kind of contours joining. However, some important drawbacks also exist. The main limitation of watershed transform is the over-segmentation due to its sensitive to noise; even the very thin noise will lead to a lot of scattered and meaningless regions. Therefore, the improvement of the watershed algorithm to deal with the over-segmentation problem becomes an essential issue.The mathematical morphology has many good traits such as effective positioning, accurate segmentation, good anti-noise performance etc. It is used in watershed transform to reduce the effect of oversegmentation. One way to achieve smoothing is to perform a morphological opening followed by a closing. The net result of these two operations is to remove or attenuate both bright and dark artifacts of noise. But it was found through the experiment that the simple use of morphological opening and closing operation will lead Edge Information discontinuous. Furthermore if the structural element is too large, it would destroy the figure of cells and affect the accuracy of segmentation. So the morphological reconstruction is widely used and satisfactorily resolved the issue. Another approach used to control oversegmentation is based on the concept of markers. An effective method for minimizing the effect of small spatial detail is to filter the image with a smoothing filter. The marker is used to define a higher"altitude"to be the only allowed regional minima. To restrict the algorithm to operate on a single watershed that contains the marker in the particular region can solve the traditional oversegmentation.Finally, the improvements of the watershed algorithm are used to deal with the colored cell image segmentation and the segmentation for overlapped-cell image. Respectively improve a segmentation method of watershed transform based on marker to deal with colored cell images and use a new method by stage to segment the overlapped cells.According to the need of clinical diagnosis, the segmentation of cells image can be roughly divided into two categories: one is to extract the nuclear from the leukocyte; another is to seperate cells from overlapping. Each has their own characteristics. We need to deal with them seperately in the pretreatment period and use different segmentation methods and technology.The improvement segmentation method of the watershed algorithm based on the marker introduce the idea of morphological reconstruction during the pretreatment, at the same time, use a relatively new method of coercing calibration of the minimum in the watershed transform. This method has greatly improved the watershed of the segmentation algorithm. With the original method of simulation experiments, the method can be more accurate segmentation results.The method to segment the overlapped-cell in stage use different scale structural elements to pretreat the image with the help of morphological reconstruction. Adopt two different technologies. Use the simple thresholding to segment the nucleus which has clear edge. We can achieve an ideal segmentation result. Then apply the watershed algorithm to the result. Take the result as the internal marker and the watershed lines as the external marker. The external marker partition the image into regions. Simply apply the watershed segmentation algorithm to each individual region. Seen from the simulation, it shows a satisfactory result.However, due to the structure of the algorithm in the form of the element is selected based on the specific circumstances of the image, adaptive capacity relatively weak, how to improve the auto-adaptive capability of the structural elements will be selected as the direction of my study at the next step .