Extraction And Classification Study Of Feature Line In Multi-objective Scene

Visual analysis is an important topic in the field of computer vision, the main research is the theories and methods of extracting structure, position and motion information of the target object in the scene from image sequences. The feature detection has a very important role in image segmentation, computer vision and pattern recognition. It’s the first problem to solve in the image processing and pattern recognition problems. Images have many features, the line features is undoubtedly an important clue to human visual perception. It can be said that at this stage of image processing and machine vision, the first step is the basic processing line feature detection, the line structure information of the target object boundary morphology can be reserved through line feature detection, thereby greatly reducing the amount of data to image processing and simplified the image analysis. Therefore, detection of feature lines has important significance and practical value.This paper studies the edge detection method for image pre-processing and feature extraction line involved, through in-depth research on mathematical morphology low cap hat transform, hot hat transform and template matching, completing the two ways of combining image feature extraction line. And through the image texture features’ difference complete classification of different target feature line.Firstly, take into account a variety of factors will be affected when the noise generated image acquisition, image filtering is necessary to improve the image quality, this paper studies the method of image pre-processing and common edge detection. Through the introduction and experiments of common methods, experiments and select the best approach to this article.Secondly, completed template matching methods integrated with the level cap transform to extract feature lines, the level cap by morphological transformation to extract the image information are peaks and valleys, and then use the template to extract the four directions of characteristics in different directions, and finally can be drawn correctly completed the characteristic line.Finally, construct a characterized feature of each pixel that on characteristic lines, then output different results according the different value of characterizing feature, the classification of different target characteristic lines will be completed., the experimental results show that this method can classify different initial completion target feature lines.