The Research Of Line Detection Algorithm Based On The Matlab

With the rapid development of computer technology, the digital imageanalysis technology as a tool to get the image information, has been widely used inagricultural, medical, military and other fields, especially the image recognitionsystem in driving the machinery. Such as the pre-warning system of line detectionof the helicopter, for the consideration of safety, the demands on the accuracy ofthe line detection is very high. Compare to the identification ability of human eyes,the computer through the analysis and processing of the image data to indentifycould collect a more scientific result within a very short time. In this process, therehave two main steps: the first is depth from defocus for the wire; the second one isthe detection and identification for the line under the complex background.This dissertation is focus on the detection for the line; therefore the main taskof this thesis is about the depth form defocus and the detection and identificationfor the electric power line. The research is divided into three parts.1. In this dissertation, it use the point spread function of Gaussian functionmodel to restore the defocusing vague images, under this situation, this studypresent a point spread function to estimate the value of fuzzy parameter, it is alsonamed as hierarchical method. After that, it evaluates the restored image using theselected clarity evaluation function. According to evaluation results to get thevalue of fuzzy parameter when the evaluation functions at maximum and to obtainthe point spread function. This process is the foundation for the ranging work.2. This dissertation has compared several clarity evaluation function in theautomatic focusing process, because this task is focus on the detection of theelectric power line, therefore, based on the discussion of a conventional method,this dissertation puts forward a new definition evaluation method: the area method, and it also tests the feasibility of the method.3. This dissertation presented a line detection algorithm based on gradientoperator. Firstly is the image grayscale transformation, using sobel operator toprocess the line detection, according to the line detection template to collect thegradient value of each pixel and then to obtain the image histogram and gradientvector histogram. Lastly it uses the matlab simulation experiment to verify thefeasibility of this algorithm, has better anti-interference and positioning accuracy,and get the more complete data and details for the image analysis.