| Abstract:Image detection technology is the foundation of recognition, segmentation, intensification and compress of mage, which is one of the important subjects in the field of computer vision and image measuring. At the first, in this paper, we introduce some traditional detection algorithms. And discuss the representative algorithms of image edge detection, such as Roberts operator、Sobel operator、Laplacian operator、 Log operator、Canny operator、Wavelet operator、Mathematical morphology operator and so on. The simulation result figures of the traditional algorithms are given and through which, we analyze the advantages and disadvantages.Then, as to the drawbacks of the traditional algorithms, in this paper, a new adaptive edge detection algorithm based on multi-scale elements is proposed which gives a new way to calculate the structure element weights. The characteristic of adaptive algorithm is that the weights of four direction structural elements can be determined by image itself, in another word, the structural elements are selected by image itself.Then, compared to the traditional algorithms, the experiment results show that the proposed algorithm can efficiently and successfully overcome the defect of weak anti-interference ability and have a better way to finish the weak edge detection of noisy image. Not only does it have good detection flexibility and extract rich edge information, but also the algorithm programming is easy to realize.Finally, we discuss the practical application of the new algorithm, which is applied to the oil slick infrared aerial image and red blood cell counting, and the red blood cell counting steps include image smoothing, histogram threshold processing, holes filling, edge detection, corrosion, refinement, modification and counting. The experimental results show that not only does the new algorithm extract clear edges, but also it can reduce the number of missed edge and false edge during edge detection processing. |
PDF Full Text Request