Target Target Edge Detection Technique Based On Neural Network Research

A SOBEL edge operator and an edge detection algorithm based on CP neural network was employed in this thesis, to achieve image processing of target digital images collecting from static explosion test of fragment warhead, and exactly detecting and analyzing of target fragments numbers, and also the ratio of perforation of fragments from the test, through which more accurate tactics performance characters can be obtained for warhead lethality assessment. SOBEL edge operator was used to extract the edge contours of two-valued gray level image as the raw sample data, simplifying the test data, and reserving the original characteristics of test data, discontinuity of perforation boundary. The edge detection approach based on CP neural network is valid to overcome the randomness of fragment dispersing and complexity of distribution of perforation holes. Applying neural network approach, the influence of the negative factors is reduced in artificial detection and statistics. In this way, the target image was detected exactly and successfully.In this thesis, the neural network and edge detection technologies were reviewed firstly, then discussed the application in weapon engineering, and illuminated the features and significances of the study. Secondly, the main algorithms and its application relate to edge detection were summarized, and then the statistics theory and target edge detection based on neural network was discussed in detail, referred to fragment warhead. Moreover, the edge detection method based on CP neural network model and technologies was given, through which the target images were detected exactly. Finally, the detection method based on neural network and artificial detection method for target detecting were contrasted and summarized, that demonstrated the feasibility and superiority of the neural network method.In a word, the experiment data statistics efficiency and accuracy was improved considerably, employing the special neural network technique to target detection in fragment warhead perforation test. The technique is applicable for target detections of a variety of fragment warhead test. Keywords: Neural Network, Target, Edge Detection, Fragment Warhead...