编织视觉叙事：超越原子视觉匹配的智能体图像束组合

摘要

图像检索传统上被建模为逐点匹配问题，即对每个候选图像进行独立评分。然而，这种原子化范式未能捕捉个人照片集中人类搜索意图的复杂性——用户往往寻求由结构关系约束的紧凑视觉故事，而非孤立的快照。为解决这一局限，我们提出**图像束组合（Image Bundle Composition, IBC）**，一种新范式，将目标从对单张图像排序转变为从海量非结构化照片池中动态组合连贯的图像束。由于目标束并非预先定义，IBC 带来了严重的组合爆炸挑战，并要求对不可分解的联合相关性进行建模。为建立这一范式，我们构建了 **IBCBench**——首个包含 109,467 张图像和 667 个经核实查询的 IBC 基准数据集，并通过半自动验证流水线构建。此外，我们提出 **BundleWeaver**，一种将 IBC 重新表述为查询条件增量超边发现的智能体框架。通过利用大语言模型自适应地搜索缺失的关系角色，并利用视觉-语言模型进行整体束验证，BundleWeaver 能有效导航组合空间。大量实验表明，尽管最先进的嵌入模型和静态分解-重排范式存在关系盲区，BundleWeaver 仍取得了显著的性能提升，凸显了从原子化评分转向动态关系组合的必要性。我们的数据集和代码已公开。

English

Image retrieval has traditionally been formulated as a point-wise matching problem, where each candidate image is scored in isolation. However, this atomic paradigm fails to capture the complexity of human search intent within personal photo collections, where users often seek compact visual stories bound by structural relations rather than isolated snapshots. To address this limitation, we introduce **Image Bundle Composition (IBC)**, a novel paradigm that shifts the objective from ranking individual images to dynamically composing cohesive image bundles from a massive, unstructured photo pool. Since target bundles are not predefined, IBC presents a severe combinatorial explosion challenge and demands modeling non-decomposable joint relevance. To establish this paradigm, we construct **IBCBench**, the first IBC benchmark dataset containing 109,467 images and 667 verified queries, built via a semi-automated verification pipeline. Furthermore, we propose **BundleWeaver**, an agentic framework that reformulates IBC as query-conditioned incremental hyperedge discovery. By employing a Large Language Model to adaptively search for missing relational roles and utilizing a Vision-Language Model for whole-bundle verification, BundleWeaver effectively navigates the combinatorial space. Extensive experiments demonstrate that while state-of-the-art embedding models and static decompose-and-rerank paradigms suffer from relational blindness, BundleWeaver achieves substantial performance gains, highlighting the necessity of shifting from atomic scoring to dynamic relational composition. Our dataset and code are available.