A survey and evaluation of edge detection operators application to medical images

摘要

One of the objectives of the image analysis is to extract its dominating informations Thus we use the segmentation as a technical of tempting to associate a stamp to each pixel according to the carried information (gray level or color) and its specific distribution in the image. Thereby, the segmentation of image is defined as being the low level step of processing that extracts and describes present significant objects in a scene, the most often in the form of regions or edges... However, this operation has never been esthetical neither admit for the visual context but it's essential in different domains and various applications. For example, we can mention the medical domain (in order to detect tumors and to localize calcifications), the biomedical field (the control of the cells size for the cancer tracking...). In the literature, different methods have been elaborated in order to detect the image's edges. They are gathered in two families: on the one hand methods privileging an approach by border (derivative, surfacics, and morphological methods...) named edge approach; on the other hand those privileging an approach by regions (Markovian and Structural methods). In this work, we'll be interested in the different methods using the edge approach for the images segmentation. Many image segmentation techniques are available. Some of these techniques use the classic masks in order to detect edges in the image. These methods are known under "deriving methods" (Robert, Sobel, Prewitt, Kirsch, Robinson, Marr and Hildreth). An other family of methods of segmentation based on the maximisation of criteria in order to obtain a sort of filter that permit the distinction of the Image edges This family is the "methods of optimal filtering " (Canny, Deriche, Spacek, Petrou Kittler...). These methods differ in their edge quality providing. This quality is in. general related to the type of image processes This work is a synthesis to study these methods, we will try to describe these methods and to highlight their advantages and drawbacks In the first paragraph we describes derivative methods, In the second paragraph we are interested in methods by optimal filtering. The third paragraph will linger to other methods of edge detection methods of edge detection, especially on the method of Mondestino and Fries, the methods of snakes and methods of segmentation for color images The presentation of each of these methods will be followed by some consultations regrouping its advantages, its disadvantages, the processing's time required and an interpretation of the edge quality obtained from different edge detection operator"' towards different noise types This evaluation will play an important role for the users of applications that need the edge detection domain especially in the medical field where segmentation play an important role in the diagnostic As a conclusion, this work allows to establish a library of image segmentation using the edge approach, containing different methods and their application fields according to their performances.