| As one of the most significant characteristics of object’s shape, edge plays an important role in image processing. The research on how to extract edges of the object from complex electronchemiluminescence(ECL) images accurately has been a hot topic in image processing for a long time. Despite the fact that many algorithms can be used for edge detection, few of them can be applied to ECL images. As an important part of the analytical instrument called REFP-1, the analytical technique of ECL images can’t detect the target object’s edges completely and obtain the target’s luminance accurately. In oder to improve the above shortcoming of the analytical instruments, a relevant module is designed and implemented in this thesis.Conventional algorithms are good choices for edge detection under the conditions of clear edges and obvious contrast in the image. However, they can’t be used to extract complete, continuous, and closed boundaries from complex ECL images with uneven grayscale and noises. Furthermore, the target part to be analyzed should be segemented from ECL images firstly. Therefore, a new edge detection algorithm called SCGVF based on wavelet transform, Canny and GVF is proposed.Firstly, the development of edge detection technologies and the hardware platform of the REFP-1 used in the project are introduced.Secondly, for the analytical requirements of the target shape, the image segmentation theories based on area and boundary are discussed in detail. Meanwhile, the region growing algorithm and classical edge detection algorithms are tested for ECL images. The results show the region growing algorithm can separate the target portion effectively from the images with uniform grayscale and clear boundaries and that the classical algorithms can also detect the edges roughly. But the detected edges of classical algorithms are usually notcontinuous and have numerous pseudo edges, which can’t meet the separation requirements of the target objects.Thirdly, this thesis focuses on Canny edge detection algorithm based on the concavity analysis of the gradient histogram, the wavelet transform technique and the GVF deformation model. Then the SCGVF algorithm is proposed by taking these three technical superiorities into consideration. In this algorithm, wavelet transform is used for image preprocessing, the improved Canny as the basis of edge map and the GVF model for boundary extraction. The test result of the algorithm shows that the algorithm has good performance on shape extraction, which can meet the pratical needs. Besides, a statistical method for emission intensity of the target portion is described briefly according to the result mentioned above.Finally, a simple and easy-to-use graphical user interface is designed for pratical use of the algorithm. The features and the operation method of the GUI are describled subsequently. Then the application is packaged to get rid of the dependence on execution environments. Eventually, it is applied to the ECL analytical instruments, which enhances the analysis function of the instruments. |
