Research On Contour Feature Extraction And Local Curvature Calculation

In the research field of image processing and pattern recognition, and the expression of curve of cognition is an indispensable part in the outline of the object recognition, at the same time curvature is understanding shape feature is the most important source of information. In actual production application, high precision detection technology is the curvature necessary precision device quality.In a straight line, angle, bending is used to identify the object shape feature element, bending is an important factor to reflect the shape difference, in the small scope relative, previous bending forming angle, in a wide range of relative internal slow bending can be approximated as linear. In the past research, for the mean curvature detection part outline in a large range have proposed many methods, however, contour line of gradual change curvature of these methods, not smooth contours of the detection effect. So, how to realize the local curvature of the contour line of the accurate detection, how to solve the distinction of arc and the angle becomes a new topic.According to the saying all above, I choose research on feature extraction of contour line and local curvature calculation as the subject of my thesis. In this thesis, we propose a method to calculate the curvature of the edge segment by using the fixed length segment along the contour. This method can be applied to the accurate estimation of the local curvature. This method can be applied to any geometric image and can provide accurate estimation of the local curvature. In this method, we first use the detection characteristic of the edge operator which based on the simple cell model to detect the edge position, through the chain algorithm to track the formation of a set of continuous edge of the contour line. The edge operator is shown to have high accuracy selectivity in azimuth angle, and the possibility of achieving high accuracy curvature calculation by detected azimuth angle is discussed. Finally, divide the contour line into end-to-end edge segments, piecewise identify whether it constitutes a part of the arc or straight line, and detect their arc curvature and the convex and concave characteristics.