<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[小秋在路上~]]></title><description><![CDATA[Good Luck!]]></description><link>https://blog.youkuaiyun.com/Miracle0_0</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; Miracle0_0]]></copyright><item><title><![CDATA[Python随机生成20个自编车牌（前两个是字母，后三个是数字）]]></title><link>https://blog.youkuaiyun.com/Miracle0_0/article/details/143269548</link><guid>https://blog.youkuaiyun.com/Miracle0_0/article/details/143269548</guid><author>Miracle0_0</author><pubDate>Sun, 27 Oct 2024 15:35:49 +0800</pubDate><description><![CDATA[【代码】Python随机生成20个自编车牌（前两个是字母，后三个是数字）]]></description><category></category></item><item><title><![CDATA[（阅读笔记CVIU2024）Multimodel fore-/background alignment for seam-based parallax-tolerant image stitching]]></title><link>https://blog.youkuaiyun.com/Miracle0_0/article/details/138966588</link><guid>https://blog.youkuaiyun.com/Miracle0_0/article/details/138966588</guid><author>Miracle0_0</author><pubDate>Thu, 16 May 2024 16:40:07 +0800</pubDate><description><![CDATA[大视差图像拼接是一个具有挑战性的计算机视觉问题。尽管提出了现有的基于接缝的方法来获得令人满意的结果，但是仍然会发生诸如对象错位、消失和重复之类的问题。针对这些问题，本文提出了一种新的基于接缝的视差容差图像拼接方法，该方法利用多个扭曲模型精确对齐背景和前景区域. 为了基于深度变化对象的特征对应性估计各种空间平滑模型，引入了一种迭代算法，该算法通过为数据分配权重来选择内点并求解网格扭曲模型。此外，我们在选择和分组未对齐特征对的基础上构造前景像素的匹配置信度，从而惩罚接缝切割的重复。]]></description><category></category></item><item><title><![CDATA[（阅读笔记arXiv2023）Seam-guided local alignment and stitching for large parallax images]]></title><link>https://blog.youkuaiyun.com/Miracle0_0/article/details/138696893</link><guid>https://blog.youkuaiyun.com/Miracle0_0/article/details/138696893</guid><author>Miracle0_0</author><pubDate>Sat, 11 May 2024 15:16:12 +0800</pubDate><description><![CDATA[在图像拼接的合成步骤中，特别是对于具有视差的图像，接缝切割方法已经被证明是有效的。然而，接缝切割的有效性通常依赖于图像的粗略对齐，使得图像中存在一个局部区域，在该区域内可以找到看似合理的接缝，对于视差较大的图像，现有的对齐方法往往达不到预期的效果。本文提出了一种以接缝质量评价为指导的局部对齐和拼接方法. 首先，利用现有的图像对齐和接缝分割方法计算出初始接缝，并对接缝处的像素质量进行评价。然后，对于低质量的像素，在对齐图像中分离其封闭块，并通过SIFT流提取修改的稠密对应来局部对齐它们。]]></description><category></category></item><item><title><![CDATA[（阅读笔记 arXiv2023）Parallax-Tolerant Image Stitching with Epipolar Displacement]]></title><link>https://blog.youkuaiyun.com/Miracle0_0/article/details/137239473</link><guid>https://blog.youkuaiyun.com/Miracle0_0/article/details/137239473</guid><author>Miracle0_0</author><pubDate>Mon, 01 Apr 2024 17:07:12 +0800</pubDate><description><![CDATA[Yu, J., Yu, Y., & Da, F. (2023). Parallax-Tolerant Image Stitching with Epipolar Displacement Field. arXiv preprint arXiv:2311.16637.]]></description><category></category></item><item><title><![CDATA[学习次模函数-第2章 定义]]></title><link>https://blog.youkuaiyun.com/Miracle0_0/article/details/136987533</link><guid>https://blog.youkuaiyun.com/Miracle0_0/article/details/136987533</guid><author>Miracle0_0</author><pubDate>Sun, 24 Mar 2024 14:51:21 +0800</pubDate><description><![CDATA[次模函数可以由几个等价的性质来定义，我们现在就来介绍。加法测度是集合函数的第一个例子，其中基数是最简单的例子。众所周知的基数性质是，对于任意两个集合，则，这推广到所有的加法测度。当且仅当，前面的等式对于的所有子集一个集合函数是次模的当且仅当，对于所有的子集，我们有：。注意，如果一个函数是次模的，并且使得（我们总是假设），则对于任意两个不相交的集合，则，即： 次模性意味着次可加性（但反之则不成立）。如前所述，次模函数的最简单的例子是基数（例如，是。]]></description><category></category></item><item><title><![CDATA[学习次模函数-第1章 引言]]></title><link>https://blog.youkuaiyun.com/Miracle0_0/article/details/136984599</link><guid>https://blog.youkuaiyun.com/Miracle0_0/article/details/136984599</guid><author>Miracle0_0</author><pubDate>Sun, 24 Mar 2024 11:30:39 +0800</pubDate><description><![CDATA[在这本专著中，次模函数的理论以一种独立的方式呈现，所有结果都是从机器学习中常见的凸分析的第一原理证明的，而不是依赖于组合优化和传统的理论计算机科学概念，如拟阵或流（见，例如， [72]有关这些方法的参考书）。此外，我们提出的算法是基于传统的凸优化算法，如单纯形法线性规划，二次规划的有效集方法，椭球方法，切割平面，和条件梯度。，在计算机科学和应用数学的许多领域中有应用，例如机器学习[125，157，117，124]，计算机视觉[31，96]，运筹学[98，182]，电气网络[162]或经济学[203]。]]></description><category></category></item><item><title><![CDATA[阅读笔记（MM2023）Learning pixel-wise alignment for unsupervised image stitching]]></title><link>https://blog.youkuaiyun.com/Miracle0_0/article/details/136968176</link><guid>https://blog.youkuaiyun.com/Miracle0_0/article/details/136968176</guid><author>Miracle0_0</author><pubDate>Sat, 23 Mar 2024 16:17:20 +0800</pubDate><description><![CDATA[图像拼接旨在对同一视图中的一对图像进行对齐。对于图像拼接来说，生成具有自然结构的精确对齐是一个挑战，因为在非共面的实际场景中，没有更宽视场图像作为参考。在本文中，我们提出了一个无监督图像拼接框架，突破了单应性估计中的共面约束，实现了在有限重叠区域下的精确像素级对齐。首先，我们通过迭代密集特征匹配结合误差控制策略来生成全局变换，以减轻由大视差引入的差异。]]></description><category></category></item><item><title><![CDATA[阅读笔记（ICIP2023）Rectangular-Output Image Stitching]]></title><link>https://blog.youkuaiyun.com/Miracle0_0/article/details/136965537</link><guid>https://blog.youkuaiyun.com/Miracle0_0/article/details/136965537</guid><author>Miracle0_0</author><pubDate>Sat, 23 Mar 2024 13:29:01 +0800</pubDate><description><![CDATA[图像拼接的目的是将两幅视场重叠的图像进行拼接，以扩大视场（FoV）。然而，现有的拼接方法拼接的图像不规则，需要进行矩形化处理，耗时且容易出现不自然的现象。本文提出了第一个端到端框架--矩形输出深度图像拼接网络（RDISNet），该框架可以将两幅图像直接拼接成一幅标准的矩形图像，同时学习图像对之间的颜色一致性并保持内容的真实性。为了进一步保留拼接图像中大对象的结构，我们设计了一个扩张的BN-RCU块来扩展RDISNet的感受野，以提取丰富的空间上下文。]]></description><category></category></item><item><title><![CDATA[阅读笔记（arXiv2022）Submodularity In Machine Learning and Artificial Intelligence]]></title><link>https://blog.youkuaiyun.com/Miracle0_0/article/details/136886465</link><guid>https://blog.youkuaiyun.com/Miracle0_0/article/details/136886465</guid><author>Miracle0_0</author><pubDate>Wed, 20 Mar 2024 19:49:13 +0800</pubDate><description><![CDATA[在本文中，我们提供了次模性和超模性及其属性的简要回顾。我们提供了大量次模性定义；许多示例次模函数及其推广的完整描述；离散约束的示例；基本算法的讨论，包括最大化、最小化和其他操作；对连续次模扩展的简要概述；以及一些历史应用。然后我们转向次模性在机器学习和人工智能中的用途，包括总结，我们提供了在自然语言处理中的抽取式和抽象式总结、数据蒸馏和凝聚、数据子集和特征选择之间的差异和共性的完整说明。]]></description><category></category></item><item><title><![CDATA[阅读笔记（CVPR2020）Warping Residual Based Image Stitching for Large Parallax]]></title><link>https://blog.youkuaiyun.com/Miracle0_0/article/details/136855495</link><guid>https://blog.youkuaiyun.com/Miracle0_0/article/details/136855495</guid><author>Miracle0_0</author><pubDate>Tue, 19 Mar 2024 21:10:47 +0800</pubDate><description><![CDATA[图像拼接技术将在不同观看位置处捕获的两个图像对齐到单个较宽图像上。当捕获的3D场景不是平面的并且相机基线大时，两个图像表现出视差，其中场景结构的相对位置与每个视图非常不同。现有的图像拼接方法往往无法对视差较大的图像进行拼接。为此，提出了一种基于扭曲残差概念的图像拼接算法。我们首先估计多个单应性，并找到它们在两个图像之间的内部特征匹配。然后，我们评估每个特征匹配相对于多个单应性的翘曲残差。]]></description><category></category></item><item><title><![CDATA[阅读笔记（NPL2023）Deep learning on image stitching with multi-viewpoint images: A survey]]></title><link>https://blog.youkuaiyun.com/Miracle0_0/article/details/136211058</link><guid>https://blog.youkuaiyun.com/Miracle0_0/article/details/136211058</guid><author>Miracle0_0</author><pubDate>Wed, 21 Feb 2024 14:28:41 +0800</pubDate><description><![CDATA[Yan N, Mei Y, Xu L, et al. Deep learning on image stitching with multi-viewpoint images: A survey[J]. Neural Processing Letters, 2023: 1-36.]]></description><category></category></item><item><title><![CDATA[阅读笔记（TMM2022）Image stitching with manifold optimization]]></title><link>https://blog.youkuaiyun.com/Miracle0_0/article/details/136193623</link><guid>https://blog.youkuaiyun.com/Miracle0_0/article/details/136193623</guid><author>Miracle0_0</author><pubDate>Tue, 20 Feb 2024 16:35:30 +0800</pubDate><description><![CDATA[Zhang L, Huang H. Image stitching with manifold optimization[J]. IEEE Transactions on Multimedia, 2022.]]></description><category></category></item><item><title><![CDATA[阅读笔记（TMM2021）Natural image stitching with layered warping constraint]]></title><link>https://blog.youkuaiyun.com/Miracle0_0/article/details/136193552</link><guid>https://blog.youkuaiyun.com/Miracle0_0/article/details/136193552</guid><author>Miracle0_0</author><pubDate>Tue, 20 Feb 2024 16:33:14 +0800</pubDate><description><![CDATA[Zhang Z, Yang X, Xu C. Natural image stitching with layered warping constraint[J]. IEEE Transactions on Multimedia, 2021.]]></description><category></category></item><item><title><![CDATA[阅读笔记（IET-IP2021）Image stitching method by multi‐feature constrained alignment and colour adjustment]]></title><link>https://blog.youkuaiyun.com/Miracle0_0/article/details/136193502</link><guid>https://blog.youkuaiyun.com/Miracle0_0/article/details/136193502</guid><author>Miracle0_0</author><pubDate>Tue, 20 Feb 2024 16:31:35 +0800</pubDate><description><![CDATA[Yuan X, Zheng Y, Zhao W, et al. Image stitching method by multi‐feature constrained alignment and colour adjustment[J]. IET Image Processing, 2021, 15(7): 1499-1507.]]></description><category></category></item><item><title><![CDATA[阅读笔记（MTA2021） Image stitching by feature positioning and seam elimination]]></title><link>https://blog.youkuaiyun.com/Miracle0_0/article/details/136193470</link><guid>https://blog.youkuaiyun.com/Miracle0_0/article/details/136193470</guid><author>Miracle0_0</author><pubDate>Tue, 20 Feb 2024 16:30:27 +0800</pubDate><description><![CDATA[Qin Y, Li J, Jiang P, et al. Image stitching by feature positioning and seam elimination[J]. Multimedia Tools and Applications, 2021, 80: 20869-20881.]]></description><category></category></item><item><title><![CDATA[阅读笔记（CSVT2021）Stable linear structures and seam measurements for parallax image stitching]]></title><link>https://blog.youkuaiyun.com/Miracle0_0/article/details/136193425</link><guid>https://blog.youkuaiyun.com/Miracle0_0/article/details/136193425</guid><author>Miracle0_0</author><pubDate>Tue, 20 Feb 2024 16:29:23 +0800</pubDate><description><![CDATA[Xue W, Xie W, Zhang Y, et al. Stable linear structures and seam measurements for parallax image stitching[J]. IEEE transactions on circuits and systems for video technology, 2021, 32(1): 253-261.]]></description><category></category></item><item><title><![CDATA[阅读笔记（PR2021）Image stitching based on angle-consistent warping]]></title><link>https://blog.youkuaiyun.com/Miracle0_0/article/details/136193389</link><guid>https://blog.youkuaiyun.com/Miracle0_0/article/details/136193389</guid><author>Miracle0_0</author><pubDate>Tue, 20 Feb 2024 16:28:29 +0800</pubDate><description><![CDATA[Chen Y, Zheng H, Ma Y, et al. Image stitching based on angle-consistent warping[J]. Pattern Recognition, 2021, 117: 107993.]]></description><category></category></item><item><title><![CDATA[阅读笔记（CVPR2021）Leveraging line-point consistence to preserve structures for wide parallax image stitc]]></title><link>https://blog.youkuaiyun.com/Miracle0_0/article/details/136193367</link><guid>https://blog.youkuaiyun.com/Miracle0_0/article/details/136193367</guid><author>Miracle0_0</author><pubDate>Tue, 20 Feb 2024 16:27:44 +0800</pubDate><description><![CDATA[Jia Q, Li Z J, Fan X, et al. Leveraging line-point consistence to preserve structures for wide parallax image stitching[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2021: 12186-12195.]]></description><category></category></item><item><title><![CDATA[阅读笔记（PAA2021）A robust method for image stitching]]></title><link>https://blog.youkuaiyun.com/Miracle0_0/article/details/136193338</link><guid>https://blog.youkuaiyun.com/Miracle0_0/article/details/136193338</guid><author>Miracle0_0</author><pubDate>Tue, 20 Feb 2024 16:26:48 +0800</pubDate><description><![CDATA[Pellikka M, Lahtinen V. A robust method for image stitching[J]. Pattern Analysis and Applications, 2021, 24(4): 1847-1858.]]></description><category></category></item><item><title><![CDATA[阅读笔记（TIP2021）Image stitching based on semantic planar region consensus]]></title><link>https://blog.youkuaiyun.com/Miracle0_0/article/details/136193296</link><guid>https://blog.youkuaiyun.com/Miracle0_0/article/details/136193296</guid><author>Miracle0_0</author><pubDate>Tue, 20 Feb 2024 16:25:45 +0800</pubDate><description><![CDATA[Li A, Guo J, Guo Y. Image stitching based on semantic planar region consensus[J]. IEEE Transactions on Image Processing, 2021, 30: 5545-5558.]]></description><category></category></item></channel></rss>