IMAGINATION项目:面向广播与工业检测的图像传感器、相机架构及成像系统前沿进展

IMAGINATION Project: Advancing Image Sensors, Camera Architectures and Imaging Systems for Broadcast and Industrial Inspection

AENEAS by Editor 2026-02-10 09:56 Original
摘要
欧洲IMAGINATION项目在图像传感器、光学与相机架构及高速成像系统方面取得重大进展,主要应用于电视直播和半导体检测。项目由Caeleste、代尔夫特理工大学等工业和研究机构组成的联盟完成,推出了支持双倍慢速摄影和扩展动态范围的新型双传感器相机架构,以及用于半导体检测的每秒100亿像素高速成像平台。这些成果提升了广播工作流的图像质量和生产效率,并为半导体制造提供了更快速可靠的检测技术,同时强化了欧洲在广播和成像技术领域的产业地位。

IMAGINATION项目取得重大进展,推动广播与工业检测成像技术革新

由欧洲领先工业及研究机构组成的联盟近期完成了IMAGINATION Penta-Euripides²项目,在图像传感器、光学与相机架构以及高速成像系统领域实现了重大突破。该项目成果主要面向电视直播广播和半导体检测机器视觉等高要求应用场景。

项目联盟成员包括Caeleste、代尔夫特理工大学、Euresys、Grass Valley Nederland BV、KLA、Melexis NV、Teledyne Adimec Advanced Image Systems BV以及intoPIX SA。该联盟成功将前沿研究转化为经过验证、接近市场化的技术。

广播领域:新型相机架构与增强成像

为应对日益增长的媒体消费和日趋复杂的制作流程,IMAGINATION项目通过相机制造商、压缩专家和系统架构师的紧密合作,在广播成像领域引入了多项重要创新:

* 采用光学白光分光器的新型双传感器相机架构,实现了双倍速慢动作(从6倍提升至12倍)以及至少增加两档光圈的扩展动态范围。

* 演示了从单幅图像生成深度图的技术,为视频处理提供了基于软件的景深控制和新的创意选项。

* 成功在广播相机中实现了ticoRAW压缩,能够在保持视觉质量的同时,实现向云端和边缘处理环境的高效数据传输。

这些进展提升了直播广播工作流程的图像质量、灵活性和生产效率,并为下一代支持云制作的相机做好了准备。

半导体检测:高速机器视觉应对先进封装

针对半导体元件和先进封装日益增长的复杂性,IMAGINATION项目在检测速度、精度和数据处理方面取得关键成果:

* 通过构建涵盖相机、图像采集卡和视觉处理平台的完整100 Gb/s生态系统,将持续成像性能提升一倍,达到每秒100亿像素。

* 新平台支持以高帧率进行超高分辨率成像(高达1亿像素),能够检测实现100%全检所需的极微小缺陷。

* 演示了包括集成嵌入式图像处理的飞行时间法在内的3D成像能力,支持可靠的绝对距离测量。

* 通过先进压缩技术及全球首次在CoaXPress上实现通用数据容器(GenDC),实现了更智能的数据处理,能够在整个检测链中标准化传输图像、元数据和辅助数据。

这些成果使得在人工智能、高性能计算和先进封装驱动的半导体制造环境中,能够进行更快、更可靠的检测。

图像传感器:提升从可见光到红外的灵敏度

项目同时推进了CMOS图像传感器技术,弥合了研究与工业部署之间的差距:

* 新的像素概念(包括单光子雪崩二极管和浮置传输门像素)改善了低光性能和光子计数性能。

* 一种CMOS兼容的层交换技术展示了从可见光到短波红外范围灵敏度提升的能力。

* 新型高速、低成本传感器测试系统降低了表征成本,支持未来的传感器开发。

这些进步为空间与地球观测、汽车传感与激光雷达、科学仪器、医疗及生命科学成像等领域带来了新机遇。

产业与社会影响

IMAGINATION项目巩固了欧洲在广播技术、半导体检测和成像组件领域的地位,并带来以下贡献:

* 通过更快、更精确的检测系统,提高工业质量和生产效率。

* 借助支持云端和远程的广播工作流程,减少环境影响。

* 通过参与CoaXPress、SMPTE和AMWA-NMOS等标准化活动,促进长期生态系统发展。

该项目已产生多项专利、演示系统,并对国际标准做出了贡献。多家合作伙伴将继续在后续的Xecs项目中开展合作,确保IMAGINATION的成果能够转化为未来的产品并进入市场。

Penta和Euripides²是由AENEAS运营的尤里卡集群(Eureka Clusters)。

Summary
The IMAGINATION Penta-Euripides² project, led by a European consortium including Caeleste, Delft University of Technology, and KLA, delivered major advances in imaging technologies for broadcast and semiconductor inspection. Key outcomes include a new two-sensor camera architecture enabling higher slow-motion speeds and dynamic range for broadcasting, as well as a high-speed 10 Gpixels/second imaging ecosystem for detecting microscopic defects in semiconductor manufacturing. These innovations enhance production quality and efficiency while strengthening Europe's position in the imaging technology value chain.

The IMAGINATION Penta-Euripides² project, a major European consortium initiative, has delivered significant technological advances in imaging systems tailored for live broadcast and high-speed industrial inspection. The project united key players across the imaging value chain, including Caeleste, Delft University of Technology, Euresys, Grass Valley, KLA, Melexis, Teledyne Adimec, and intoPIX, to translate advanced research into near-market-ready technologies.

For broadcast applications, the consortium introduced several innovations to meet demands for higher quality and more complex production workflows. A novel two-sensor camera architecture, using an optical white-light splitter, enables both double-speed slow motion (increasing from 6× to 12×) and an extended dynamic range of at least two additional F-stops. The project also demonstrated software-based depth-of-field control from single-image depth maps and successfully integrated ticoRAW compression into a broadcast camera for efficient, quality-preserving transmission to cloud and edge processing environments. These developments aim to enhance image quality, creative flexibility, and productivity for next-generation, cloud-enabled production.

In the domain of semiconductor inspection, IMAGINATION tackled the challenges posed by increasingly complex components and advanced packaging. The project doubled sustained imaging performance to 10 gigapixels per second, supported by a complete 100 Gb/s ecosystem encompassing cameras, frame grabbers, and vision processors. New platforms enable very high-resolution imaging (up to 100 megapixels) at high frame rates for detecting extremely small defects, crucial for 100% inspection. The consortium also demonstrated 3D imaging capabilities, including Time-of-Flight with embedded processing, and implemented smarter data handling via advanced compression and the world's first use of the Generic Data Container (GenDC) over CoaXPress. This allows standardized transport of images, metadata, and auxiliary data, supporting faster, AI-driven inspection.

The project also pushed CMOS image sensor technology forward. Advancements include new pixel concepts like Single-Photon Avalanche Diodes (SPADs) and Floating Transfer Gate pixels for improved low-light and photon-counting performance, as well as a CMOS-compatible layer-exchange technology to boost sensitivity from the visible into the short-wave infrared range. New high-speed, low-cost sensor test systems were developed to reduce characterization costs. These sensor advances create opportunities in fields ranging from space observation and automotive LiDAR to scientific and medical imaging.

IMAGINATION strengthens Europe's industrial position in broadcast tech, semiconductor inspection, and imaging components. Its broader impact includes boosting industrial quality and productivity through faster inspection, reducing the environmental footprint of broadcast via cloud workflows, and supporting long-term ecosystem development through contributions to standards like CoaXPress, SMPTE, and AMWA-NMOS. The project has generated patents, demonstrators, and standard contributions, with several partners continuing collaboration in follow-up Xecs projects to commercialize the results.

The project was conducted under the Penta and Euripides² Eureka Clusters, operated by AENEAS.

Résumé
Le projet européen IMAGINATION, regroupant des acteurs industriels et académiques comme Caeleste, Grass Valley, KLA et intoPIX, a abouti à des avancées technologiques majeures dans deux secteurs. Pour la diffusion télévisuelle en direct, il a développé une nouvelle architecture de caméra à deux capteurs permettant un ralenti extrême et une plage dynamique étendue, ainsi que la compression ticoRAW pour le traitement cloud. Pour l'inspection des semi-conducteurs, le projet a doublé les performances d'imagerie à 10 Gpixels/seconde et introduit le standard GenDC sur CoaXPress pour gérer de très gros volumes de données, permettant ainsi des contrôles plus rapides et précis.

The IMAGINATION Penta-Euripides² project delivered major advances in image sensors, optical and camera architectures, and high-speed imaging systems, addressing demanding applications in live television broadcasting and machine vision for semiconductor inspection. The project was carried out by a European consortium bringing together leading industrial and research players across the imaging value chain, including Caeleste, Delft University of Technology, Euresys, Grass Valley Nederland BV, KLA, Melexis NV, Teledyne Adimec Advanced Image Systems BV and intoPIX SA.

Together, the consortium successfully translated advanced research into validated, near-market technologies.

New camera architectures and enhanced imaging for broadcast

To support growing media consumption and increasingly complex production workflows, IMAGINATION introduced significant innovations in broadcast imaging, driven by close collaboration between camera manufacturers, compression specialists and system architects:

A new two-sensor camera architecture using an optical white-light splitter enabled both double-speed slow motion (from 6× to 12×) and an extended dynamic range of at least two additional F-stops.

Depth maps generated from single images were demonstrated, enabling software-based depth-of-field control and new creative options in video processing.

ticoRAW compression was successfully implemented in a broadcast camera, enabling efficient transmission to cloud and edge processing environments while preserving visual quality.

These developments improve image quality, flexibility and productivity in live broadcast workflows and prepare next-generation cameras for cloud-enabled production.

High-speed machine vision for advanced semiconductor packaging

IMAGINATION addressed the rapidly increasing complexity of semiconductor components and advanced packaging, where inspection speed, accuracy and data handling are critical:

Sustained imaging performance was doubled to 10 Gpixels per second, enabled by a complete 100 Gb/s ecosystem spanning cameras, frame grabbers and vision processing platforms.

New platforms support very high-resolution imaging (up to 100 megapixels) at high frame rates, enabling detection of extremely small defects required for 100% inspection.

3D imaging capabilities, including Time-of-Flight approaches with embedded image processing, were demonstrated, supporting reliable absolute distance measurements.

Smarter data handling was achieved through advanced compression and the world’s first implementation of the Generic Data Container (GenDC) over CoaXPress, enabling standardized transport of images, metadata and auxiliary data across the inspection chain.

These results enable faster, more reliable inspection in semiconductor manufacturing environments driven by AI, HPC and advanced packaging.

More sensitive image sensors from visible to infrared

The project also advanced CMOS image sensor technologies, bridging the gap between research and industrial deployment:

New pixel concepts improved low-light and photon-counting performance, including Single-Photon Avalanche Diodes (SPADs) and Floating Transfer Gate (FTG) pixels.

A CMOS-compatible layer-exchange technology demonstrated increased sensitivity from the visible into the short-wave infrared range.

New high-speed, low-cost sensor test systems reduced characterization costs and support future sensor development.

These advances open opportunities in space and Earth observation, automotive sensing and LiDAR, scientific instrumentation, medical and life-science imaging.

Industrial, societal and ecosystem impact

IMAGINATION strengthens Europe’s position in broadcast technology, semiconductor inspection and imaging components, while also contributing to:

Higher industrial quality and productivity through faster and more accurate inspection systems

Reduced environmental impact via cloud-enabled and remote broadcast workflows

Long-term ecosystem development through standardisation activities, including CoaXPress, SMPTE and AMWA-NMOS

The project resulted in patents, demonstrators and contributions to international standards, and several partners continue collaboration in follow-up Xecs projects, ensuring that IMAGINATION’s results are carried forward into future products and markets.

To learn more about the project, its scope and outcomes, see the official Project Impact Summary.

Penta and Euripides² are Eureka Clusters, operated by AENEAS.

The post IMAGINATION Project: Advancing Image Sensors, Camera Architectures and Imaging Systems for Broadcast and Industrial Inspection appeared first on Aeneas.

AI Insight
Core Point

A European consortium delivered advanced imaging technologies for broadcast and semiconductor inspection, translating research into near-market innovations.

Key Players

Caeleste — image sensor design, Belgium.

Delft University of Technology — research, Netherlands.

Euresys — frame grabbers/vision software, Belgium.

Grass Valley Nederland BV — broadcast equipment, Netherlands.

KLA — semiconductor inspection, USA.

Melexis NV — sensors, Belgium.

Teledyne Adimec — industrial cameras, Netherlands.

intoPIX SA — compression IP, Belgium.

Industry Impact
  • ICT: High — new compression and cloud-enabled broadcast workflows.
  • Terminals/Consumer Electronics: Medium — improved camera architectures for media production.
  • Computing/AI: High — enables AI-driven high-speed semiconductor inspection.
  • Automotive: Low — sensor advances may benefit LiDAR and sensing.
Tracking

Strongly track — project outputs directly enable next-gen semiconductor inspection and cloud-based broadcast production.

Highlights
Tech Breakthrough
Categories
半导体 人工智能 科研
AI Processing
2026-04-03 23:06
deepseek / deepseek-chat