MLMei Liu

Mei Liu

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Mei Liu has a total of 7 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Sub-pixel detection of screw tooth profile based on improved Zernike moments

    by Yang Xu, Xiao Wang, Peisi Zhong, Yanzhi Bi, Mei Liu, Weiwei Ge

    In order to improve the measurement accuracy of ball screw geometry, a sub-pixel detection method of ball screw contour based on improved Zernike moments is proposed. Firstly, the acquired ball screw image is smoothed by median filtering, and then the Canny operator is selected to coarsely locate the edges of the tooth contour to extract the edge pixels of the target area where the contour of the ball screw to be measured is located. On the basis of Ghosal algorithm, an improved Zernike algorithm is proposed to obtain the sub-pixel edge coordinates of the image. After experiments, it is shown that the detection method has the advantages of high accuracy and good stability for the positioning of sub-pixel edge coordinates.


  • Open Access

    Tomato grading based on Machine vision D-S multi-feature fusion algorithm

    by Yanzhi Bi, Chao Zhang, Peisi Zhong, Yang Xu, Mei Liu, Shufang Zhou

    Aiming at the problem that the detection speed is slow and the classification standard cannot be comprehensively rated when grading tomato quality at present, this paper proposes an improved Sobel edge detection algorithm and multi-feature fusion classification model. Firstly, by combining fast median filtering and morphological operation method, and improving the calculation of the gradient direction of Sobel operator. It can better extract the edge image features of tomato, and provide better image information for subsequent size, maturity and defect detection. Then, a multi-feature fusion algorithm based on D-S evidence theory was proposed. The algorithm process carried out the classification results by D-S evidence theory to calculate a new belief function to assign different weights, and realized a multi-feature fusion classification model. Finally, through the experimental verification, the online detection speed reaches 6 /s, and the average accuracy of tomato quality classification detection is 91%. Compared with the single feature classification, it is found that the multi-feature fusion classification is relatively high in speed and accuracy, which realizes the rapid detection and comprehensive classification of tomato quality classification.