The power of fusion: LiDAR meets hyperspectral imaging in a new era of exploration

LiDAR systems have developed as a pivotal technology in the contemporary landscape, revolutionizing various fields ranging from autonomous vehicles to environmental monitoring and urban planning [1,2]. These systems operate by emitting laser pulses and determining the time it takes for them to return after bouncing off objects, thereby creating highly accurate three-dimensional representations of the surrounding environment [3]. The significance of LiDAR in modern times lies in its remarkable ability to capture detailed spatial data with precision and speed, supporting highly accurate analysis and deeper insights [4,5]. From enhancing the safety and efficiency of transportation networks to aiding in conserving natural resources and cultural heritage sites, LiDAR systems have become indispensable tools for researchers, engineers, and policymakers alike. As advancements continue to refine the technology and expand its applications, the influence of LiDAR on shaping the future of various industries and scientific endeavours only continues to grow [6].

LiDAR is a vital tool thanks to its exceptional ability to provide highly accurate and comprehensive spatial data, supporting a wide range of applications. In contrast to conventional surveying methods, which often entail significant time and labour investments, LiDAR enables swift data acquisition over extensive areas with unmatched accuracy. This capability holds immense importance in diverse tasks such as urban planning terrain mapping, ecological studies involving monitoring changes in forest canopy height, and the development of intricate 3D models for infrastructure projects. Furthermore, LiDAR's unique ability to penetrate dense foliage and adverse atmospheric conditions renders it particularly invaluable in fields like forestry, agriculture, and disaster management [[7], [8], [9]]. Its pivotal role extends to cutting-edge technologies like autonomous vehicles, where its precise mapping functionalities are fundamental for navigation and obstacle avoidance [10]. In essence, LiDAR's necessity stems from its provision of detailed spatial information, essential for addressing multifaceted challenges across various domains, thereby propelling innovation and advancement in countless sectors [11].

While LiDAR excels in delivering high-resolution three-dimensional structural information, it lacks the spectral richness needed to capture biochemical and material properties. Hyperspectral imaging (HSI), by contrast, provides fine-grained spectral detail that enables identification of vegetation health, mineral composition, and surface materials, but typically at lower spatial resolution [12,13]. When combined, LiDAR and HSI offer synergistic benefits: improved land cover classification, enhanced detection of vegetation stress, and more comprehensive analysis of environmental and geological features. However, integration also presents challenges [13,14]. Differences in spatial and spectral resolution, acquisition geometry, and sensor alignment often introduce registration errors and geometric mismatches. Additionally, the high dimensionality of HSI combined with dense LiDAR point clouds increases computational complexity and the risk of overfitting in machine learning models, particularly when annotated training datasets are limited. These trade-offs underscore the need for careful fusion strategies and set the stage for exploring both the promise and limitations of LiDAR-HSI integration [15,16]. These considerations make it essential to evaluate fusion strategies carefully, and this review aims to highlight both the current capabilities and the future prospects of LiDAR–HSI integration.

While conventional LiDAR provides precise 3D structural information, it lacks the spectral richness needed to identify biochemical or material properties. Integrating LiDAR with HSI bridges this gap by combining geometric accuracy with detailed spectral signatures, enabling more accurate land-cover classification, early vegetation stress detection, and comprehensive environmental analysis. The evolution of LiDAR systems marks a fascinating journey of technological advancement and innovation. Originating in the early 1960s primarily for military applications, LiDAR systems have undergone significant transformations over the decades. Initially bulky and expensive, early LiDAR systems were limited in their capabilities and largely confined to specialized use cases. However, with advancements in laser technology, computing power, and miniaturization, LiDAR systems have become more compact, affordable, and versatile. Table 1 provides information on the evolution of LiDAR systems over time [17]. The integration of solid-state lasers, photodetectors, and sophisticated algorithms has enabled LiDAR to evolve into a mainstream technology with widespread applications across industries. Moreover, developments in software algorithms have enhanced data processing capabilities, allowing for more efficient and accurate interpretation of LiDAR-generated 3D point clouds (PCs). As LiDAR continues to evolve, ongoing research and development efforts focus on improving resolution, increasing range, reducing costs, and expanding functionality, thereby unlocking new possibilities for its use in fields such as autonomous vehicles, urban planning, environmental monitoring, and archaeology [[18], [19], [20]]. The most advanced development in current LiDAR technology is hyperspectral LiDAR [21]. This technology was introduced to address the challenges of integrating LiDAR with hyperspectral imagery and to eliminate matching errors that typically occur during the fusion process. Hyperspectral LiDAR can simultaneously acquire a three-dimensional hyperspectral point cloud, combining both structural and spectral information of the target. This allows for precise characterization of the target’s three-dimensional physiological and biochemical parameters [22]. Moreover, because all spectral bands share a unified emission and reception channel, it fundamentally avoids the geometric mismatches that arise when separately aligning LiDAR data with hyperspectral images.

The paper is structured as follows: Section 2 delves into the operational mechanisms of LiDAR systems. Section 3 comprehensively explores the various types and attributes of LiDAR systems, encompassing airborne, terrestrial, mobile, and spaceborne platforms. In Section 4, a detailed examination of the applications of both LiDAR systems and hyperspectral imaging (HSI) is presented. Section 5 addresses the processing methods of HSI data, including pre-processing techniques, empirical relationships, radiative transfer modelling, and applications of machine learning and deep learning. Section 6 elucidates the challenges and prospects inherent in LiDAR systems. Finally, concluding remarks are provided in Section 7 to wrap up the discussion.