首页 > AI前沿 > Sharp First-Hit Identification from Finite Depth Samples

Sharp First-Hit Identification from Finite Depth Samples

arXiv机器学习 2026-11-15 02:49 12 阅读 查看原文
用户现在需要把一段英文新闻正文排版成语义化的HTML片段,遵循给定的规则。首先,先看内容,整个是一段?不对,先梳理结构,首先开头的设问,然后研究内容,然后方法,然后推导结果,然后实验和结论?不对,按照规则,自然段落用p,长文可以加h2小节对吧?首先,首先有没有可以划分的小节?比如首先问题的提出,然后研究方法,然后理论推导结果,然后实验验证,最后结论?不对,要保证不增删内容,哦不对,不能增内容,那小节标题?不对哦,规则说不得增删改内容,那h2/h3是划分小节,但是标题不能自己加?哦不对,看原文的内容,原文是连贯的,那是不是整个内容是一个大的,然后有没有需要划分的?不对,再仔细看规则:保留原文所有信息,不得增删改内容,所以h2/h3的标题不能自己瞎编,哦不对,原文里没有小节标题,那是不是整个正文是一个段落?不对,等下看原文的逻辑: 第一部分是提出问题:“Which depths in a new camera view are determined by finitely many source-depth measurements? Pointwise uncertainty bounds do not answer this question: unobserved geometry between samples can change which surface is encountered first.” 这部分是问题引入。 然后是研究内容:“We study a bounded Lipschitz class of continuous source inverse-depth fields with interval observations and known camera pose. Building on classical Lipschitz envelopes, we characterize the exact set of attainable first-hit depths and construct a globally admissible scene for every value in that set. The set can be disconnected. We also give a necessary and sufficient test for several proposed depths to be first hits of one shared scene.” 这部分是研究方法和初步结果。 然后是理论推导:“An epipolar reduction uses scalar cone inequalities without enumerating surfaces. In a rectified model with source spacing $h$, interval half-width $\varepsilon$, baseline $b$, and slope bound $L$, we prove a sharp worst-case inverse-depth diameter of $\max\{Lh+2\varepsilon,2\varepsilon/(1-bL)\}$ when $bL<1$, subject to sufficient depth and domain range for attainment. Exact samples still give the sharp bound $Lh$ at $bL=1$, while above this threshold ambiguity need not vanish as sampling becomes arbitrarily dense. Any positive measurement uncertainty can destroy uniform stability at the critical threshold.” 这部分是核心推导。 然后是实验和结论:“We provide constructive examples, an exact rational validation oracle, and a frozen ten-scene evaluation using reference, stereo, and learned-depth observations. The experiments separate solver correctness under the scene assumptions from failure of those assumptions: fixed learned-depth bands often yield narrow but incorrect sets. The results identify what a specified scene class permits; they do not infer a continuum regularity or physical-scene guarantee from sampled predictions.” 哦,但是规则说不能增删改内容,所以小节标题不能自己加?那是不是直接把每个自然段用p?不对,原文是一个连续的文本,没有分段?不对,用户给的正文是一段?不对,看用户给的正文,是从开头到结尾是一整段?不对,等下看用户给的正文: 用户给的正文是: Which depths in a new camera view are determined by finitely many source-depth measurements? Pointwise uncertainty bounds do not answer this question: unobserved geometry between samples can change which surface is encountered first. We study a bounded Lipschitz class of continuous source inverse-depth fields with interval observations and known camera pose. Building on classical Lipschitz envelopes, we characterize the exact set of attainable first-hit depths and construct a globally admissible scene for every value in that set. The set can be disconnected. We also give a necessary and sufficient test for several proposed depths to be first hits of one shared scene. An epipolar reduction uses scalar cone inequalities without enumerating surfaces. In a rectified model with source spacing $h$, interval half-width $\varepsilon$, baseline $b$, and slope bound $L$, we prove a sharp worst-case inverse-depth diameter of $\max\{Lh+2\varepsilon,2\varepsilon/(1-bL)\}$ when $bL<1$, subject to sufficient depth and domain range for attainment. Exact samples still give the sharp bound $Lh$ at $bL=1$, while above this threshold ambiguity need not vanish as sampling becomes arbitrarily dense. Any positive measurement uncertainty can destroy uniform stability at the critical threshold. We provide constructive examples, an exact rational validation oracle, and a frozen ten-scene evaluation using reference, stereo, and learned-depth observations. The experiments separate solver correctness under the scene assumptions from failure of those assumptions: fixed learned-depth bands often yield narrow but incorrect sets. The results identify what a specified scene class permits; they do not infer a continuum regularity or physical-scene guarantee from sampled predictions. 哦,原来用户给的是一整段?不对,那要不要划分?不对,规则说“长文用

/

划分小节”,但是如果原文没有小节标题的话,不能自己加,对吧?因为不能增内容。那是不是整个内容用一个p?不对,不对,再看,有没有重要的句子可以用strong?比如核心的结论?比如“We prove a sharp worst-case inverse-depth diameter of $\max\{Lh+2\varepsilon,2\varepsilon/(1-bL)\}$ when $bL<1$”还有“Exact samples still give the sharp bound $Lh$ at $bL=1$”还有“Any positive measurement uncertainty can destroy uniform stability at the critical threshold”还有最后的结论? 不对,首先,不能增删改内容,所以所有的文字都要保留,包括公式的$符号?对,原文的公式是带$的,要保留。然后,自然段落,原文如果是一个完整的自然段,那用一个p?不对,等下,有没有可能把逻辑上的部分用h2?但是h2需要有内容啊,不能自己加标题,那不行,因为那属于新增内容了。哦对哦,规则说不得增删改内容,所以不能自己加小节标题,所以h2/h3这里如果没有原文自带的小标题的话,就不能用?不对,再仔细看规则:“长文用

/

划分小节”,但是如果原文没有小节的话,那就不用?那现在原文是一段,那就用一个p?不对,不对,再想想,有没有可能我理解错了?不对,再看内容,有没有引用?没有。有没有列表?没有。有没有代码?没有。那只有p,还有重要的句子用strong? 等下,再梳理重要的句子: 1. 问题部分的开头?不对,核心的研究结果,比如“we characterize the exact set of attainable first-hit depths and construct a globally admissible scene for every value in that set”?不对,还有证明的直径公式那里,是核心结论,可以加strong。还有“Any positive measurement uncertainty can destroy uniform stability at the critical threshold”也是重要结论。还有最后“The results identify what a specified scene class permits; they do not infer a continuum regularity or physical-scene guarantee from sampled predictions”也是重要结论。 不对,但是要注意不要改动内容,只是加strong标签。那现在,整个正文是一段,用p包裹,然后把重要的句子用strong? 等下,再检查规则:1.只输出HTML片段,不要Markdown,不要解释。2.每个自然段落用p包裹,长文用h2/h3划分小节,但是这里没有小节标题,所以不用h2/h3?因为没有原文的小节,自己加的话就是新增内容了,不符合规则4。3.列表用ul,引用用blockquote,代码用pre,这里都没有,所以不用。重要句子用strong。4.保留所有信息,不增删改。 那现在,组织内容:

Which depths in a new camera view are determined by finitely many source-depth measurements? Pointwise uncertainty bounds do not answer this question: unobserved geometry between samples can change which surface is encountered first. We study a bounded Lipschitz class of continuous source inverse-depth fields with interval observations and known camera pose. Building on classical Lipschitz envelopes, we characterize the exact set of attainable first-hit depths and construct a globally admissible scene for every value in that set. The set can be disconnected. We also give a necessary and sufficient test for several proposed depths to be first hits of one shared scene. An epipolar reduction uses scalar cone inequalities without enumerating surfaces. In a rectified model with source spacing $h$, interval half-width $\varepsilon$, baseline $b$, and slope bound $L$, we prove a sharp worst-case inverse-depth diameter of $\max\{Lh+2\varepsilon,2\varepsilon/(1-bL)\}$ when $bL<1$, subject to sufficient depth and domain range for attainment. Exact samples