Papers by Keyword: Line Detection

Abstract: Automatic recognition of the line in the image is an important work in the field of machine vision and image processing. Focusing on the problem of the computational cost and large invalid sampling in the line extraction algorithm using standard Hough transform (HT). An improved HT algorithm is proposed to solve these problems. The parameters of the improved algorithm can be reduced to one and the accumulator is operated by setting the tolerance. Then the existence of linear is determined by seting the threshold. The experimental results show that the algorithm not only can effectively solve the problem of local maxima and improves the algorithm speed and reduces the storage space,but also the accuracy of line extraction is improved.

Abstract: Hough Transform ^[ has become a common method in the usage of line detection because of its robustness. It is important in computer vision and image analysis. Usually, the standard Hough transform method (SHT) transform the points in image space into parameter space and vote for all the possible patterns passing through that point. But, there are two serious problems in the standard method of line detection. The first is the high computation complexity and the second is the large storage requirements .In order to solve the two problems, this paper raise a fast-Hough transform algorithm base on pyramid algorithm. First of all we need to desample the primitive binary image with n times; and execute the Hough transform in the nth level image to get the parameter of straight line in this image, which is used in the n-1 level image. Finally we can get the parameter of lines in the primitive image. Experiments show that this method can extremely reduces the computational time.

Abstract: Fast processing and recognition of image have an important influence on the real-time performance of AGV by vision navigation. The fast method of edge detection based on mathematical morphology for straight line path is proposed in this paper. Binary processing is implemented firstly for path image. Then the edge detection using morphological gradient can be executed directly, and image filtering and recognition of path edge can be realized simultaneously. Kinematics model of AGV is built and the shape of structure element in morphology processing can be adjusted in real time. The results of experiment show that recognition time of straight line path is shorten by this image processing method.

Abstract: The concept of Statistical Directional Characteristics (SDCs) is introduced in this paper. The applications of SDCs in Synthetic Aperture Radar (SAR) imagery interpretation are discussed. Two applications, i.e., target recognition and line detection, are raised to demonstrate the effectiveness of the SDCs-based methods. Both numerical and graphical results are presented. According to the experiment results, the SDCs-based methodology shows advantages both in robustness and in efficiency.

Abstract: Fast and accurate acquisition of navigation information is the key and premise for robot guidance. In this paper, a robot trajectory guidance system composed of a camera, a Digital Signal Controller and mobile agency driven by stepper motors is given. First the JPEG (Joint Photographic Expert Group) image taken by camera is decoded and turns to correspond pixel image. By binarization process the image is then transformed to a binary image. A fast line extraction algorithm is presented based on Column Elementary Line Segment method. Furthermore the trajectory direction deviation parameters and distance deviation parameters are calculated. In this way the robot is controlled to follow the given track accurately in higher speed.

Abstract: In view of the inherent limitations of classical Hough transform in the detection of line, an improved algorithm of randomized Hough transform is offered in this paper. The feature point sets are segmented based on the connected component labels in this algorithm. The elements of various point sets are stored in sequence so as to reducing the invalid samples. By using a valid line detection area and a dynamic storage management, the temporary storage units are gradually reducing with the iterative process. The speed of the line detection and the accuracy of identification can be improved, the storage space can be reduced, the computational complexity can be cut down and some unnecessary calculations can be avoided by using this algorithm. However, it still has the characteristics of the classical Hough transform that it is not sensitive to noise. The line edge where the cutting tool is on can be pinpointed. To achieve the vision-based mode of tool wear state, it can accurately detect rake and flank face of the cutting tool, and can effectively remove the influence to the tool wear detection from the BUILT-UP EDGE in cutting process. By the validation of testing examples, the algorithm has high speed, low memory occupation and high accuracy.