www.ptreview.co.uk
21
'26
Written on Modified on
Ultra-compact hyperspectral sensing with AI
Fraunhofer IPMS unveils a compact AI-based hyperspectral camera that enables real-time material and quality analysis for industrial and agricultural applications.
www.fraunhofer.de
Display of spectral image in the field © Fraunhofer IPMS
The OASYS project focuses on optoelectronic sensors for application-oriented systems. In the subproject A1 a team of scientists and industry experts is developing an ultra-compact, energy-efficient hyperspectral camera that uses artificial intelligence to perform complex material and quality analyses in real time. The integrated spectrometer records spectral characteristics, revealing chemical properties that are invisible to the human eye. This enables defects in food or the composition of textiles or plastics to be identified quickly and accurately.
The novel hyperspectral camera offers wide-ranging applications in industrial and agricultural processes. Its innovative approach combines conventional 2D imaging with artificial intelligence and spectral analysis. A standard 2D camera first captures a high-resolution image of the target object. Artificial intelligence then analyzes the image in real time, automatically identifying regions of interest. The integrated spectrometer subsequently performs spectral analysis exclusively at these selected positions, determining the chemical composition. This intelligent approach significantly enhances the efficiency of hyperspectral measurements. Rather than capturing spectral data across the entire image - a computationally intensive process - the system analyzes only the relevant measurement points. This targeted method substantially reduces data volumes, energy consumption, and processing time.
Heinrich Engelke at work: Determining textile types using an ultra-compact intelligent hyperspectral camera © BTU / Fraunhofer IPMS
The information obtained in this way supports, for example, the reliable sorting of textiles and plastics. It also increases the reliability of identifying counterfeit products. Additionally, it improves quality control in food processing by detecting pressure marks and defects and enables an accurate assessment of plant condition and nutrient requirements in agriculture. Automated evaluation enables faster and more reliable decision-making. At the same time, processes become more sustainable, and economic resources are used more efficiently.
The components developed in the project will form the basis for future sensor systems that could significantly improve industry, recycling, agriculture, and the food sector.
The OASYS project focuses on optoelectronic sensors for application-oriented systems. In the subproject A1 a team of scientists and industry experts is developing an ultra-compact, energy-efficient hyperspectral camera that uses artificial intelligence to perform complex material and quality analyses in real time. The integrated spectrometer records spectral characteristics, revealing chemical properties that are invisible to the human eye. This enables defects in food or the composition of textiles or plastics to be identified quickly and accurately.
The novel hyperspectral camera offers wide-ranging applications in industrial and agricultural processes. Its innovative approach combines conventional 2D imaging with artificial intelligence and spectral analysis. A standard 2D camera first captures a high-resolution image of the target object. Artificial intelligence then analyzes the image in real time, automatically identifying regions of interest. The integrated spectrometer subsequently performs spectral analysis exclusively at these selected positions, determining the chemical composition. This intelligent approach significantly enhances the efficiency of hyperspectral measurements. Rather than capturing spectral data across the entire image - a computationally intensive process - the system analyzes only the relevant measurement points. This targeted method substantially reduces data volumes, energy consumption, and processing time.
Heinrich Engelke at work: Determining textile types using an ultra-compact intelligent hyperspectral camera © BTU / Fraunhofer IPMS
The information obtained in this way supports, for example, the reliable sorting of textiles and plastics. It also increases the reliability of identifying counterfeit products. Additionally, it improves quality control in food processing by detecting pressure marks and defects and enables an accurate assessment of plant condition and nutrient requirements in agriculture. Automated evaluation enables faster and more reliable decision-making. At the same time, processes become more sustainable, and economic resources are used more efficiently.
The components developed in the project will form the basis for future sensor systems that could significantly improve industry, recycling, agriculture, and the food sector.
www.ipms.fraunhofer.de
