Infrared Image Edge Detection Based on Morphology-Canny Fusion Algorithm

Tang Qingju, Liu Yuanlin, Bu Chiwu, Li Dayong
2016 International Journal of u- and e- Service, Science and Technology  
Effectively extracting the contour edge of the defect in the infrared image can realize the recognition of the geometrical features of the defect. In view of the traditional Canny edge detection algorithm in the Gauss filter variance and high and low threshold selection need artificial intervention, does not have the adaptive ability, and its defects in the gradient calculation, proposed a method based on the improved Canny operator and image morphological fusion edge detection method. Using
more » ... improved Canny operator and image morphology to edge detection, the simulation results show that the algorithm has good anti noise ability, effectively improve the accuracy and integrity of the edge detection in the infrared image, and achieve a better extraction of geometric features of the edge of the defect. the interference of noise to be missed, the noise point for the detection of edge points out. At the same time, the edge of the image is blurred, and the edge is not high. So the effect is not ideal in the actual image processing. In 1986, Canny John proposed that the edge detection operator should satisfy three criteria. First, signal to noise ratio criterion; Second, positioning accuracy criteria; Third, single edge response criterion, and the optimal edge detection operator -Canny operator is derived. Compared to the differential operators, the Canny operator based on the optimization algorithm is widely used in the high SNR and high detection accuracy, and has become the standard for evaluating other edge detection methods. But the variance of the Gauss function and the choice of the high and low threshold are all manual setting, and the adaptive ability is poor; moreover, it uses the finite difference mean value of 2×2 neighborhood, and is sensitive to noise [11] [12] . Mathematical morphology is a new discipline which is based on image analysis. It can be used to measure and extract the shape of the image in order to realize the image analysis and recognition. It has obvious advantages in solving the problem of removing noise and edge detection. Basic operations include Expansion, corrosion, open operation and close operation. Due to the continuous improvement of the edge detection algorithm, the integration of various algorithms and algorithms has become a new trend in the current research. The combining mathematical morphology and Canny operator edge extraction method, using morphology open operator to estimate the background, and original image global gray of geometric computing effect to reduce the background complex texture and noise of the image, to extract the defect geometric feature of edges better.
doi:10.14257/ijunesst.2016.9.3.25 fatcat:64lbwavcyjcgppunpjjznk7eom