The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
Para medição inteligente da visão, a extração de características geométricas da imagem é uma questão essencial. Primitivo de interesse de contorno (CPI) significa um recurso de contorno de formato regular situado em um objeto alvo, que é amplamente utilizado para cálculo geométrico em medição de visão e servoacionamento. Para perceber que o modelo de extração CPI pode ser aplicado de forma flexível a diferentes objetos novos, a extração CPI baseada em aprendizagem única pode ser implementada com rede neural convolucional profunda, usando apenas uma imagem de suporte anotada para orientar o processo de extração CPI. Neste artigo, propomos uma rede de extração de primitivas de interesse de contorno de vários estágios (MS-CPieNet), que usa a estratégia de vários estágios para melhorar a capacidade de discriminação de CPI e antecedentes complexos. Em segundo lugar, o módulo de atenção espacial não local é utilizado para aprimorar as características profundas, fundindo globalmente as características da imagem com alcances curtos e longos. Além disso, a classificação densa em 4 direções é projetada para obter a direção normal do contorno, e as direções podem ser usadas posteriormente para o pós-processo de desbaste do contorno. A eficácia dos métodos propostos é validada pelos experimentos com os conjuntos de dados OCP e ROCM. Experimentos de medição 2-D são conduzidos para demonstrar a aplicação conveniente do MS-CPieNet proposto.
Jinyan LU
Henan University of Engineering
Quanzhen HUANG
Henan University of Engineering
Shoubing LIU
Henan University of Engineering
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copiar
Jinyan LU, Quanzhen HUANG, Shoubing LIU, "Multi-Stage Contour Primitive of Interest Extraction Network with Dense Direction Classification" in IEICE TRANSACTIONS on Information,
vol. E105-D, no. 10, pp. 1743-1750, October 2022, doi: 10.1587/transinf.2022EDP7031.
Abstract: For intelligent vision measurement, the geometric image feature extraction is an essential issue. Contour primitive of interest (CPI) means a regular-shaped contour feature lying on a target object, which is widely used for geometric calculation in vision measurement and servoing. To realize that the CPI extraction model can be flexibly applied to different novel objects, the one-shot learning based CPI extraction can be implemented with deep convolutional neural network, by using only one annotated support image to guide the CPI extraction process. In this paper, we propose a multi-stage contour primitives of interest extraction network (MS-CPieNet), which uses the multi-stage strategy to improve the discrimination ability of CPI and complex background. Second, the spatial non-local attention module is utilized to enhance the deep features, by globally fusing the image features with both short and long ranges. Moreover, the dense 4-direction classification is designed to obtain the normal direction of the contour, and the directions can be further used for the contour thinning post-process. The effectiveness of the proposed methods is validated by the experiments with the OCP and ROCM datasets. A 2-D measurement experiments are conducted to demonstrate the convenient application of the proposed MS-CPieNet.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2022EDP7031/_p
Copiar
@ARTICLE{e105-d_10_1743,
author={Jinyan LU, Quanzhen HUANG, Shoubing LIU, },
journal={IEICE TRANSACTIONS on Information},
title={Multi-Stage Contour Primitive of Interest Extraction Network with Dense Direction Classification},
year={2022},
volume={E105-D},
number={10},
pages={1743-1750},
abstract={For intelligent vision measurement, the geometric image feature extraction is an essential issue. Contour primitive of interest (CPI) means a regular-shaped contour feature lying on a target object, which is widely used for geometric calculation in vision measurement and servoing. To realize that the CPI extraction model can be flexibly applied to different novel objects, the one-shot learning based CPI extraction can be implemented with deep convolutional neural network, by using only one annotated support image to guide the CPI extraction process. In this paper, we propose a multi-stage contour primitives of interest extraction network (MS-CPieNet), which uses the multi-stage strategy to improve the discrimination ability of CPI and complex background. Second, the spatial non-local attention module is utilized to enhance the deep features, by globally fusing the image features with both short and long ranges. Moreover, the dense 4-direction classification is designed to obtain the normal direction of the contour, and the directions can be further used for the contour thinning post-process. The effectiveness of the proposed methods is validated by the experiments with the OCP and ROCM datasets. A 2-D measurement experiments are conducted to demonstrate the convenient application of the proposed MS-CPieNet.},
keywords={},
doi={10.1587/transinf.2022EDP7031},
ISSN={1745-1361},
month={October},}
Copiar
TY - JOUR
TI - Multi-Stage Contour Primitive of Interest Extraction Network with Dense Direction Classification
T2 - IEICE TRANSACTIONS on Information
SP - 1743
EP - 1750
AU - Jinyan LU
AU - Quanzhen HUANG
AU - Shoubing LIU
PY - 2022
DO - 10.1587/transinf.2022EDP7031
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E105-D
IS - 10
JA - IEICE TRANSACTIONS on Information
Y1 - October 2022
AB - For intelligent vision measurement, the geometric image feature extraction is an essential issue. Contour primitive of interest (CPI) means a regular-shaped contour feature lying on a target object, which is widely used for geometric calculation in vision measurement and servoing. To realize that the CPI extraction model can be flexibly applied to different novel objects, the one-shot learning based CPI extraction can be implemented with deep convolutional neural network, by using only one annotated support image to guide the CPI extraction process. In this paper, we propose a multi-stage contour primitives of interest extraction network (MS-CPieNet), which uses the multi-stage strategy to improve the discrimination ability of CPI and complex background. Second, the spatial non-local attention module is utilized to enhance the deep features, by globally fusing the image features with both short and long ranges. Moreover, the dense 4-direction classification is designed to obtain the normal direction of the contour, and the directions can be further used for the contour thinning post-process. The effectiveness of the proposed methods is validated by the experiments with the OCP and ROCM datasets. A 2-D measurement experiments are conducted to demonstrate the convenient application of the proposed MS-CPieNet.
ER -