Pub Date : 2020-07-01DOI: 10.1109/ICIVC50857.2020.9177440
Jia-yi Zhang, Y. Liu, Zhi-qiang Liu
Through analysis the characteristics of micro parts such as multi-categories, high detection frequency and similar shape in the classification process, based on the FAsT_Match algorithm, an improved algorithm of template matching recognition is proposed which is a Grid Region Optimized FAsT_Match (GRO FAsT_Match for short). Firstly, the method of gray level adjustment, global threshold image segmentation, boundary tracking and denoising is used to extract the smallest rectangle of the target part image as ROI area. Secondly, by calculating the scale relationship between ROI region and template image, the step sizes and limits of grid parameters for translation and scaling transformation are optimized. In order to improve the discrimination of normalized SAD distance for similar parts, uniform sampling of template image is adopted. The experimental data show that this algorithm features fast, precise, clear distinguish of similar parts, and meets the requirements of micro parts classification and detection. It has practical significance to improve the assembly efficiency of micro parts.
{"title":"An Improved FAsT_Match Algorithm for Micro Parts Detection","authors":"Jia-yi Zhang, Y. Liu, Zhi-qiang Liu","doi":"10.1109/ICIVC50857.2020.9177440","DOIUrl":"https://doi.org/10.1109/ICIVC50857.2020.9177440","url":null,"abstract":"Through analysis the characteristics of micro parts such as multi-categories, high detection frequency and similar shape in the classification process, based on the FAsT_Match algorithm, an improved algorithm of template matching recognition is proposed which is a Grid Region Optimized FAsT_Match (GRO FAsT_Match for short). Firstly, the method of gray level adjustment, global threshold image segmentation, boundary tracking and denoising is used to extract the smallest rectangle of the target part image as ROI area. Secondly, by calculating the scale relationship between ROI region and template image, the step sizes and limits of grid parameters for translation and scaling transformation are optimized. In order to improve the discrimination of normalized SAD distance for similar parts, uniform sampling of template image is adopted. The experimental data show that this algorithm features fast, precise, clear distinguish of similar parts, and meets the requirements of micro parts classification and detection. It has practical significance to improve the assembly efficiency of micro parts.","PeriodicalId":6806,"journal":{"name":"2020 IEEE 5th International Conference on Image, Vision and Computing (ICIVC)","volume":"13 1","pages":"24-28"},"PeriodicalIF":0.0,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90464670","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-07-01DOI: 10.1109/ICIVC50857.2020.9177481
Xunwei Tong, Ruifeng Li, Lianzheng Ge, Lijun Zhao, Ke Wang
In this paper, we propose a pose refinement method based on the visible surface extraction of 3D object. Given a rough estimation of object pose, the algorithm of iterative closet point (ICP) is often used to refine the pose by aligning the object model with test scene. To avoid the interference of invisible points on the ICP process, we only use the visible surface for pose refinement. It is especially necessary when occlusion occurs in the scene. Combining the technologies of image rendering and depth consistency verification, the visible surface can be effectively extracted. During the process of pose refinement, hypothesis verification methods are also used to eliminate unreasonable hypothetical poses as early as possible. The proposed method is evaluated on the public Tejani dataset. The experimental results show that our method improved the average F1-score by 0.2062, which proves that our method can obtain pose estimation results of high accuracy, even in the occluded scene.
{"title":"Pose Refinement of Occluded 3D Objects Based on Visible Surface Extraction","authors":"Xunwei Tong, Ruifeng Li, Lianzheng Ge, Lijun Zhao, Ke Wang","doi":"10.1109/ICIVC50857.2020.9177481","DOIUrl":"https://doi.org/10.1109/ICIVC50857.2020.9177481","url":null,"abstract":"In this paper, we propose a pose refinement method based on the visible surface extraction of 3D object. Given a rough estimation of object pose, the algorithm of iterative closet point (ICP) is often used to refine the pose by aligning the object model with test scene. To avoid the interference of invisible points on the ICP process, we only use the visible surface for pose refinement. It is especially necessary when occlusion occurs in the scene. Combining the technologies of image rendering and depth consistency verification, the visible surface can be effectively extracted. During the process of pose refinement, hypothesis verification methods are also used to eliminate unreasonable hypothetical poses as early as possible. The proposed method is evaluated on the public Tejani dataset. The experimental results show that our method improved the average F1-score by 0.2062, which proves that our method can obtain pose estimation results of high accuracy, even in the occluded scene.","PeriodicalId":6806,"journal":{"name":"2020 IEEE 5th International Conference on Image, Vision and Computing (ICIVC)","volume":"34 1","pages":"176-181"},"PeriodicalIF":0.0,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76164413","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-07-01DOI: 10.1109/ICIVC50857.2020.9177441
Ke Li, L. Kong, Yifeng Zhang
In clinical practice, the determination of the location, shape, and size of brain tumor can greatly assist the diagnosis, monitoring, and treatment of brain tumor. Therefore, accurate and reliable automatic brain tumor segmentation algorithm is of great significance for clinical diagnosis and treatment. With the rapid development of deep learning technology, more and more efficient image segmentation algorithms have also been applied in this field. It has been proven that U-Net model combined with variational auto-encoder can help to effectively regularize the shared encoder and thereby improve the performance. Based on the VAE-U-Net model, this paper proposes a structure called VAE skip connection. By fusing the position information contained in VAE branch into U-Net decoding stage, the network can retain more high-resolution detail information. In addition, we integrate ShakeDrop regularization into the networks to further alleviate the overfitting problem. The experimental results show that the networks after adding VAE skip connection and ShakeDrop can achieve competitive results on the BraTS 2018 dataset.
在临床实践中，确定脑肿瘤的位置、形状和大小对脑肿瘤的诊断、监测和治疗有很大的帮助。因此，准确可靠的脑肿瘤自动分割算法对临床诊断和治疗具有重要意义。随着深度学习技术的快速发展，越来越多高效的图像分割算法也被应用于该领域。研究表明，U-Net模型结合变分自编码器可以有效地对共享编码器进行正则化，从而提高编码器的性能。基于VAE- u - net模型，提出了一种VAE跳接结构。通过将VAE支路中包含的位置信息融合到U-Net解码阶段，可以保留更多高分辨率的细节信息。此外，我们将ShakeDrop正则化集成到网络中，以进一步缓解过拟合问题。实验结果表明，加入VAE跳跃连接和ShakeDrop后的网络在BraTS 2018数据集上可以取得有竞争力的结果。
{"title":"3D U-Net Brain Tumor Segmentation Using VAE Skip Connection","authors":"Ke Li, L. Kong, Yifeng Zhang","doi":"10.1109/ICIVC50857.2020.9177441","DOIUrl":"https://doi.org/10.1109/ICIVC50857.2020.9177441","url":null,"abstract":"In clinical practice, the determination of the location, shape, and size of brain tumor can greatly assist the diagnosis, monitoring, and treatment of brain tumor. Therefore, accurate and reliable automatic brain tumor segmentation algorithm is of great significance for clinical diagnosis and treatment. With the rapid development of deep learning technology, more and more efficient image segmentation algorithms have also been applied in this field. It has been proven that U-Net model combined with variational auto-encoder can help to effectively regularize the shared encoder and thereby improve the performance. Based on the VAE-U-Net model, this paper proposes a structure called VAE skip connection. By fusing the position information contained in VAE branch into U-Net decoding stage, the network can retain more high-resolution detail information. In addition, we integrate ShakeDrop regularization into the networks to further alleviate the overfitting problem. The experimental results show that the networks after adding VAE skip connection and ShakeDrop can achieve competitive results on the BraTS 2018 dataset.","PeriodicalId":6806,"journal":{"name":"2020 IEEE 5th International Conference on Image, Vision and Computing (ICIVC)","volume":"3 1","pages":"97-101"},"PeriodicalIF":0.0,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"80228093","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-07-01DOI: 10.1109/ICIVC50857.2020.9177487
Xiong Chuan, Zhu Dandan, Peng Ziming
Virtual reality technology can provide a concurrent and multi-sensory interaction between humans and computers, enhancing the immersion, realism and fun in the process of human-computer interaction. The introduction of virtual reality technology into the design and development of fitness games can change the deficiencies of traditional fitness games such as low dimensionality, boringness, and low substitution. It makes the fitness process interesting and efficient. This article takes a VR spinning bike game designed and developed based on the Unity3D platform as an example to explore the application of VR technology in the fitness game industry.
{"title":"Design and Development of Spinning Bike Game Based on VR Technology","authors":"Xiong Chuan, Zhu Dandan, Peng Ziming","doi":"10.1109/ICIVC50857.2020.9177487","DOIUrl":"https://doi.org/10.1109/ICIVC50857.2020.9177487","url":null,"abstract":"Virtual reality technology can provide a concurrent and multi-sensory interaction between humans and computers, enhancing the immersion, realism and fun in the process of human-computer interaction. The introduction of virtual reality technology into the design and development of fitness games can change the deficiencies of traditional fitness games such as low dimensionality, boringness, and low substitution. It makes the fitness process interesting and efficient. This article takes a VR spinning bike game designed and developed based on the Unity3D platform as an example to explore the application of VR technology in the fitness game industry.","PeriodicalId":6806,"journal":{"name":"2020 IEEE 5th International Conference on Image, Vision and Computing (ICIVC)","volume":"43 1","pages":"227-231"},"PeriodicalIF":0.0,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77409415","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-07-01DOI: 10.1109/ICIVC50857.2020.9177490
Mengying Xu, Jie Zhou, Rui Yang
In recent years, cognitive radio sensor networks (CRSNs) have been commonly applied in environmental monitoring and image acquisition. However, recent advances in channel allocation have led to lower network reward, lifetime, and energy utilization rate. As a basic and fundamental problem to obtain image data in CRSNs, it governs the performance of CRSNs. To further improve the reward and throughput of obtaining image, this paper proposes an improved immune hybrid bat algorithm (IIHBA) based on bat algorithm. Furthermore, we develop a simulation environment and compared the performance of IIHBA with particle swarm optimization (PSO) and genetic algorithm (GA). Last but not the least, computational experiments showed that the reward is improved 11.36%, 27.20% respectively compared with GA and PSO when the number of users is 20 and the number of channels is 5. Based on the above findings, the proposed scheme can improve the reward of system, especially in terms of higher-throughput.
{"title":"A Biologically Inspired Channel Allocation Method for Image Acquisition in Cognitive Radio Sensor Networks","authors":"Mengying Xu, Jie Zhou, Rui Yang","doi":"10.1109/ICIVC50857.2020.9177490","DOIUrl":"https://doi.org/10.1109/ICIVC50857.2020.9177490","url":null,"abstract":"In recent years, cognitive radio sensor networks (CRSNs) have been commonly applied in environmental monitoring and image acquisition. However, recent advances in channel allocation have led to lower network reward, lifetime, and energy utilization rate. As a basic and fundamental problem to obtain image data in CRSNs, it governs the performance of CRSNs. To further improve the reward and throughput of obtaining image, this paper proposes an improved immune hybrid bat algorithm (IIHBA) based on bat algorithm. Furthermore, we develop a simulation environment and compared the performance of IIHBA with particle swarm optimization (PSO) and genetic algorithm (GA). Last but not the least, computational experiments showed that the reward is improved 11.36%, 27.20% respectively compared with GA and PSO when the number of users is 20 and the number of channels is 5. Based on the above findings, the proposed scheme can improve the reward of system, especially in terms of higher-throughput.","PeriodicalId":6806,"journal":{"name":"2020 IEEE 5th International Conference on Image, Vision and Computing (ICIVC)","volume":"12 1","pages":"267-271"},"PeriodicalIF":0.0,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81950067","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-07-01DOI: 10.1109/ICIVC50857.2020.9177477
Xianping Li
Triangular meshes have become popular in image representation and have been widely used in image processing to improve computational efficiency and accuracy. This paper introduces anisotropic mesh adaptation (AMA) to represent color images with fewer points while keeping good quality. Finite element interpolation is used in the image reconstruction from the triangular mesh-based representation. A few methods are proposed to deal with the different color channels for representation purpose. Experimental results for various images show that single triangular mesh can represent the color image as good as three-mesh representation and the differences are not significant.
{"title":"Anisotropic Mesh Representation for Color Images","authors":"Xianping Li","doi":"10.1109/ICIVC50857.2020.9177477","DOIUrl":"https://doi.org/10.1109/ICIVC50857.2020.9177477","url":null,"abstract":"Triangular meshes have become popular in image representation and have been widely used in image processing to improve computational efficiency and accuracy. This paper introduces anisotropic mesh adaptation (AMA) to represent color images with fewer points while keeping good quality. Finite element interpolation is used in the image reconstruction from the triangular mesh-based representation. A few methods are proposed to deal with the different color channels for representation purpose. Experimental results for various images show that single triangular mesh can represent the color image as good as three-mesh representation and the differences are not significant.","PeriodicalId":6806,"journal":{"name":"2020 IEEE 5th International Conference on Image, Vision and Computing (ICIVC)","volume":"43 1","pages":"139-143"},"PeriodicalIF":0.0,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79368535","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-07-01DOI: 10.1109/ICIVC50857.2020.9177453
Zhe Zheng, Lin Lei, Hao Sun, Gangyao Kuang
Object detection is an important part of remote sensing image analysis. With the development of the earth observation technology and convolutional neural network, remote sensing image object detection technology based on deep learning has received more and more attention and research. At present, many excellent object detection algorithms have been proposed and applied in the field of remote sensing. In this paper, the object detection algorithms of remote sensing image is systematically summarized, the main contents include the traditional remote sensing image object detection method and the method based on deep learning, emphasis on summarize the remote sensing image object detection algorithm based on deep learning and its development course, then we introduced the rule of performance evaluation of object detection and datasets that commonly used. Finally, the future development trend is analyzed and prospected. It is hoped that this summary and analysis can provide some reference for future research on object detection technology in remote sensing field.
{"title":"A Review of Remote Sensing Image Object Detection Algorithms Based on Deep Learning","authors":"Zhe Zheng, Lin Lei, Hao Sun, Gangyao Kuang","doi":"10.1109/ICIVC50857.2020.9177453","DOIUrl":"https://doi.org/10.1109/ICIVC50857.2020.9177453","url":null,"abstract":"Object detection is an important part of remote sensing image analysis. With the development of the earth observation technology and convolutional neural network, remote sensing image object detection technology based on deep learning has received more and more attention and research. At present, many excellent object detection algorithms have been proposed and applied in the field of remote sensing. In this paper, the object detection algorithms of remote sensing image is systematically summarized, the main contents include the traditional remote sensing image object detection method and the method based on deep learning, emphasis on summarize the remote sensing image object detection algorithm based on deep learning and its development course, then we introduced the rule of performance evaluation of object detection and datasets that commonly used. Finally, the future development trend is analyzed and prospected. It is hoped that this summary and analysis can provide some reference for future research on object detection technology in remote sensing field.","PeriodicalId":6806,"journal":{"name":"2020 IEEE 5th International Conference on Image, Vision and Computing (ICIVC)","volume":"4 1","pages":"34-43"},"PeriodicalIF":0.0,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83083681","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-07-01DOI: 10.1109/ICIVC50857.2020.9177459
Yongyuan Xue, Tianhang Gao
Visual SLAM is the technology that complete self-localization and build environment map synchronously. The feature point extraction and matching of the input image is very important for visual SLAM to achieve pose calculation and map building. For most of the literature feature point extraction and matching algorithms, the change of illumination may have a great impact on the final matching results. To address the issue, this paper proposes a novel feature point extraction and matching method based on Akaze algorithm (IICS-Akaze). Histogram equalization and dark channel prior theory are combined to construct a color space with constant illumination. Akaze algorithm is adopted for fast multi-scale feature extraction to generate feature point descriptors. The feature points are then quickly matched through the FLANN, and RANSC is introduced to eliminate the mismatches. In addition, the experiments on open data set are conducted in terms of feature extraction quantity, matching accuracy, and illumination robustness among the related methods. The experimental results show that the proposed method is able to accurately extract and match image feature points when the illumination changes dramatically.
{"title":"Feature Point Extraction and Matching Method Based on Akaze in Illumination Invariant Color Space","authors":"Yongyuan Xue, Tianhang Gao","doi":"10.1109/ICIVC50857.2020.9177459","DOIUrl":"https://doi.org/10.1109/ICIVC50857.2020.9177459","url":null,"abstract":"Visual SLAM is the technology that complete self-localization and build environment map synchronously. The feature point extraction and matching of the input image is very important for visual SLAM to achieve pose calculation and map building. For most of the literature feature point extraction and matching algorithms, the change of illumination may have a great impact on the final matching results. To address the issue, this paper proposes a novel feature point extraction and matching method based on Akaze algorithm (IICS-Akaze). Histogram equalization and dark channel prior theory are combined to construct a color space with constant illumination. Akaze algorithm is adopted for fast multi-scale feature extraction to generate feature point descriptors. The feature points are then quickly matched through the FLANN, and RANSC is introduced to eliminate the mismatches. In addition, the experiments on open data set are conducted in terms of feature extraction quantity, matching accuracy, and illumination robustness among the related methods. The experimental results show that the proposed method is able to accurately extract and match image feature points when the illumination changes dramatically.","PeriodicalId":6806,"journal":{"name":"2020 IEEE 5th International Conference on Image, Vision and Computing (ICIVC)","volume":"22 1","pages":"160-165"},"PeriodicalIF":0.0,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"75046774","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper presents an approach to single-object video object segmentation, only using the first-frame bounding box (without mask) to initialize. The proposed method is a tracking-driven single-object video object segmentation, which combines an effective Box2Segmentation module with a general object tracking module. Just initialize the first frame box, the Box2Segmentation module can obtain the segmentation results based on the predicted tracking bounding box. Evaluations on the single-object video object segmentation dataset DAVIS2016 show that the proposed method achieves a competitive performance with a Region Similarity score of 75.4% and a Contour Accuracy score of 73.1%, only under the settings of first-frame bounding box initialization. The proposed method outperforms SiamMask which is the most competitive method for video object segmentation under the same settings, with Region Similarity score by 5.2% and Contour Accuracy score by 7.8%. Compared with the semi-supervised VOS methods without online fine-tuning initialized by a first frame mask, the proposed method also achieves comparable results.
{"title":"Td-VOS: Tracking-Driven Single-Object Video Object Segmentation","authors":"Shaopan Xiong, Shengyang Li, Longxuan Kou, Weilong Guo, Zhuang Zhou, Zifei Zhao","doi":"10.1109/ICIVC50857.2020.9177471","DOIUrl":"https://doi.org/10.1109/ICIVC50857.2020.9177471","url":null,"abstract":"This paper presents an approach to single-object video object segmentation, only using the first-frame bounding box (without mask) to initialize. The proposed method is a tracking-driven single-object video object segmentation, which combines an effective Box2Segmentation module with a general object tracking module. Just initialize the first frame box, the Box2Segmentation module can obtain the segmentation results based on the predicted tracking bounding box. Evaluations on the single-object video object segmentation dataset DAVIS2016 show that the proposed method achieves a competitive performance with a Region Similarity score of 75.4% and a Contour Accuracy score of 73.1%, only under the settings of first-frame bounding box initialization. The proposed method outperforms SiamMask which is the most competitive method for video object segmentation under the same settings, with Region Similarity score by 5.2% and Contour Accuracy score by 7.8%. Compared with the semi-supervised VOS methods without online fine-tuning initialized by a first frame mask, the proposed method also achieves comparable results.","PeriodicalId":6806,"journal":{"name":"2020 IEEE 5th International Conference on Image, Vision and Computing (ICIVC)","volume":"51 1","pages":"102-107"},"PeriodicalIF":0.0,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79154877","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-07-01DOI: 10.1109/ICIVC50857.2020.9177486
Jiehua Zhang, Zhang Li, Qifeng Yu
Image registration is a basic task in biological image processing. Different stained histology images contain different clinical information, which could assist pathologists to diagnose a certain disease. It is necessary to improve the accuracy of image registration. In this paper, we present a robust registration method that consists of three steps: 1) extracting match points; 2) a pre-alignment consisting of a rigid transformation and an affine transformation on the coarse level; 3) an accurate non-rigid registration optimized by the extracted points. The existing methods use the features of the image pair to initial alignment. We proposed a new metric for the non-rigid transformation which adding the part of optimizing extracting points into the original metric. We evaluate our method on the dataset from the ANHIR Registration Challenge and use MrTRE (median relative target registration error) to measure the performance on the training data. The test result illustrates that the presented method is accurate and robust.
{"title":"Point-Based Registration for Multi-stained Histology Images","authors":"Jiehua Zhang, Zhang Li, Qifeng Yu","doi":"10.1109/ICIVC50857.2020.9177486","DOIUrl":"https://doi.org/10.1109/ICIVC50857.2020.9177486","url":null,"abstract":"Image registration is a basic task in biological image processing. Different stained histology images contain different clinical information, which could assist pathologists to diagnose a certain disease. It is necessary to improve the accuracy of image registration. In this paper, we present a robust registration method that consists of three steps: 1) extracting match points; 2) a pre-alignment consisting of a rigid transformation and an affine transformation on the coarse level; 3) an accurate non-rigid registration optimized by the extracted points. The existing methods use the features of the image pair to initial alignment. We proposed a new metric for the non-rigid transformation which adding the part of optimizing extracting points into the original metric. We evaluate our method on the dataset from the ANHIR Registration Challenge and use MrTRE (median relative target registration error) to measure the performance on the training data. The test result illustrates that the presented method is accurate and robust.","PeriodicalId":6806,"journal":{"name":"2020 IEEE 5th International Conference on Image, Vision and Computing (ICIVC)","volume":"41 1","pages":"92-96"},"PeriodicalIF":0.0,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81318867","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：