Webinar Sistemas hiperespectrales

Introduction to InfiMon and Hyper-Spectral Technology

Overview of the Session

• Carolina López introduces herself and outlines the session's focus on recent products in the market, specifically within the realm of vision systems and hyper-spectral technology.

• She expresses gratitude to attendees and her team, emphasizing a dynamic session where questions can be asked via chat for real-time engagement.

Structure of the Presentation

• The session is designed to last one hour, with an aim to cover significant information efficiently while allowing for audience inquiries on specific topics.

Company Background

• InfiMon has over 25 years of experience in artificial vision solutions and image analysis, focusing exclusively on these areas.

• The company employs a technical team that conducts visibility studies and offers commercial advice tailored to clients' needs regarding vision systems.

Diverse Offerings and Services

Multi-brand Approach

• InfiMon operates as a multi-brand company, collaborating with various manufacturers to provide optimal solutions based on client requirements.

Client Support Services

• They assist clients who may need additional help due to project overlaps or unfamiliarity with certain technologies by offering development support.

Product Catalog Insights

Comprehensive Product Range

• The catalog includes a wide array of components necessary for vision systems such as lighting, optics, cameras, software, and mounting accessories.

Market Applications

• InfiMon serves multiple sectors beyond automotive and industrial applications; they also cater to leisure, multimedia, sports-related applications, security systems, quality control processes, etc.

Personal Introduction

Professional Background

• Carolina López shares her background as a telecommunications engineer since 2006 at Infimon. She initially managed medical imaging products before leading the Technical Department since 2019.

Team Experience

• Emphasizes the maturity and extensive experience of her team members who have been part of Infimon for many years.

Session Focus: Hyper-Spectral Technology

Upcoming Topics

• A brief introduction will be provided about hyper-spectral technology followed by discussions on its applications. The presentation will conclude with an overview of recent innovations in hyper-spectral imaging available in their catalog.

Evolution of Computer Vision Technologies

Historical Development of Visual Systems

• The presentation highlights the evolution of computer vision, starting with monochrome camera systems, followed by the introduction of color cameras which opened up new application opportunities.

• The next significant technological advancement was the emergence of 3D systems, marking a pivotal shift in equipment capabilities.

Spectral Imaging and Its Maturity

• Spectral imaging is presented as a mature technology that has been available for many years, with advancements in product offerings since its inception around 2006 at InfiMoon.

• The discussion emphasizes that current spectral imaging products differ significantly from earlier versions, showcasing an evolution in capabilities and applications.

Understanding Hyperspectral Imaging

• Hyperspectral imaging applies spectroscopy to analyze how light interacts with materials based on their physical properties and molecular composition.

• A graphical representation illustrates absorption bands for common molecules like water (H2O), highlighting a notable absorption band around 1450 nanometers.

Practical Applications of Hyperspectral Imaging

• An example is provided where conventional cameras struggle to detect water levels in bottles; however, hyperspectral imaging can effectively identify these levels due to specific light absorption characteristics.

• The concept is further clarified by explaining that areas containing water absorb radiation at 1450 nanometers, resulting in darker images where water is present.

Components of Hyperspectral Cameras

• A basic breakdown of a hyperspectral camera includes a sensor and optical components known as an image spectrograph.

• The entrance slit component allows only a micrometric line of light into the spectrograph while blocking all other light, enabling spectral decomposition before projection onto the sensor.

Hyperpectral Imaging Technology Overview

Pushbroom Technology in Hyperspectral Imaging

• The pushbroom technology allows for the inspection of hyperspectral data by scanning a product linearly, requiring either camera movement or product movement to gather complete information.

• A hyperspectral camera captures only a single line of data at a time, resulting in vertical patterns in spectral images due to the nature of spectral decomposition.

Data Acquisition and Dataset Creation

• Completing the scan of a sample generates multiple spectral images, creating a dataset known as a hyperspectral cube, which contains detailed spectral information across various wavelengths.

• By viewing this hyperspectral cube from different perspectives, one can extract specific images corresponding to selected spectral bands (e.g., 670 nanometers).

Band Selection and System Limitations

• The system can capture data across numerous bands (e.g., 224), akin to using multiple selective filters; however, implementing such systems industrially is often impractical.

• An alternative solution based on muscle technique allows for real-time data acquisition as products move along conveyor belts or drones traverse fields.

Spectral Range and Applications

• Current applications utilize UVNIR technology covering wavelengths from 400 to 1000 nanometers; additional solutions exist within shortwave infrared ranges (1000 - 2500 nanometers).

• Notably, some sensors operate between 900 and 1700 nanometers for cost-effective industrial applications while maintaining sensitivity.

Cost Implications and Market Variability

• There is confusion regarding NIR definitions among manufacturers; clarity is needed when discussing equipment that operates beyond certain wavelength thresholds.

• As one moves towards longer wavelengths (medium wave infrared: 3 - 5 micrometers), costs increase significantly due to sensor technology and optical systems required for effective operation.

Summary of Key Insights

• The price variability in hyperspectral imaging systems correlates with the spectral range utilized; moving away from infrared increases costs due to advanced sensor requirements.

Applications of Hyperspectral Imaging in Quality Control

Overview of Applications

• The speaker introduces various applications of hyperspectral imaging, particularly in the food industry, emphasizing its potential for quality control and contamination detection.

Quality Control in Fruits and Vegetables

• Hyperspectral imaging can enhance quality control by detecting bruises, analyzing ripeness, firmness, and water content beyond conventional imaging techniques.

• Different fruits require tailored approaches due to variations in skin thickness; for instance, kiwis versus apples or oranges versus watermelons.

Contaminant Detection

• The technology is increasingly used to identify contaminants such as shell fragments or foreign objects (e.g., gloves or wood pieces) that may mix with products during processing.

• Differentiating between fruit and contaminants using hyperspectral analysis is feasible because it detects specific chemical components like fatty acids absent in peels.

Chemical Composition Analysis

• The method allows for a deeper understanding of the chemical makeup of products rather than just visual characteristics, providing insights into their composition through spectral analysis.

Meat Quality Control

• Hyperspectral imaging offers advantages over X-ray methods for detecting contaminants in meat by assessing surface-level quality attributes without penetrating deeply into the product.

• While not a complete substitute for X-rays due to limited penetration depth, hyperspectral imaging can analyze pH levels, sugar content, fat content, moisture levels, and salt concentration effectively.

Thermo-Sealing Control

• Recent interest has emerged around using hyperspectral techniques to monitor thermo-sealing processes by identifying obstructions that prevent proper sealing of packaging materials.

Plastic Classification Techniques

• The speaker discusses advancements in classifying plastics using hyperspectral imaging across various types (e.g., ABS, PVC), highlighting differences achieved through different spectral bands.

• Traditional NIR cameras are commonly employed for plastic classification; however, newer models provide enhanced capabilities for accurate identification based on spectral data.

Detection of Plastics Using SWIR Technology

Overview of SWIR Technology

• The Short-Wave Infrared (SWIR) system operates between 1000 and 2500 nanometers, effectively detecting various plastics, including transparent polycarbonate, which is not easily identified by other techniques like OVENIR (400-1000 nm) or MIR (900-1700 nm).

Limitations of Other Techniques

• Traditional methods struggle to clearly identify polycarbonate due to its transparency; detection relies on light orientation and reflections at the edges rather than the material itself.

• In completely smooth and transparent areas, light passes through without leaving a clear signature, making it difficult for cameras to capture accurate images of the product.

Advantages of SWIR in Plastic Identification

• SWIR provides a distinct spectral signature that enhances the ability to identify various plastics such as ABS, PVC, polystyrene, nylon, and PET. However, limitations exist for detecting transparent polycarbonate.

Specific Products for Plastic Detection

• The Fx17 model has limitations in identifying certain plastics compared to more advanced models like Fx50. The latter is specifically designed for black plastics but also performs well with polycarbonate.

Applications Beyond Plastic Detection

• The FX50 operates within 3000 to 5000 nanometers and is tailored for classifying black plastics while also being useful in quality control processes across different industries such as wood and pharmaceuticals.

Applications in Pharmaceuticals and Geology

Pharmaceutical Applications

• In pharmaceuticals, systems using short-wave infrared can classify medications through packaging materials like blisters. This allows for accurate identification even when pills are mixed.

Geology Applications

• Various spectral bands are utilized in geological applications for mineral mapping. Long Wave Infrared technology is particularly expensive but effective for aerial inspections and detailed material classification.

Conclusion: Importance of Spectral Analysis

• Different infrared technologies provide unique advantages depending on the application—whether it's plastic identification or geological analysis—highlighting the importance of selecting appropriate tools based on specific needs.

Hyper-Spectral Technology in Art Analysis

Importance of Non-Invasive Techniques

• Hyper-spectral technology is significant for analyzing paintings, focusing on non-invasive techniques that gather information without physical contact.

• The comparison between X-ray and hyper-spectral imaging highlights the ability to "strip" layers of paint, revealing underlying details not visible to the naked eye.

Applications in Restoration

• Spectral analysis aims to detect specific molecules by identifying their light signatures, which can aid in restoration efforts by identifying historical binders used in paints.

• A study from Jean Geomatics on Altamira caves illustrates ongoing research aimed at extracting more detailed information about artworks.

Enhanced Imaging Capabilities

• Hyper-spectral imaging allows for deeper insights into artwork details that may be invisible to untrained eyes, enhancing understanding for art historians and restorers.

• Current market interests include applications related to color detection and electronic recycling, particularly in identifying transparent coatings through spectral techniques.

Product Development and Evolution

• Discussion shifts towards product solutions available for various applications; collaboration with Specke has led to advancements in hyper-spectral imaging technology over the years.

• Modern products are more compact and efficient compared to earlier versions, offering improved sensitivity and speed for industrial applications.

Specific Equipment Recommendations

• Different models like FX10 (400–1000 nm), FX17 (900–1700 nm), and FX50 (2700–5300 nm) cater to various spectral needs based on application requirements.

• For pigment detection similar to studies conducted at Altamira, a WNIR system operating within 400–1000 nm is recommended depending on desired outcomes.

What is the Phoenix System?

Overview of the Phoenix System

• The Phoenix system consists of expensive and specialized equipment primarily intended for research and geological applications rather than industrial use.

Expansion of Spikim's Catalog

• Spikim has significantly expanded its catalog, now offering not just cameras but also a variety of accessories to enhance usability, allowing users to easily set up and start capturing hyperspectral images.

Accessories for Enhanced Functionality

• A range of accessories includes scanners that hold cameras from Spikim’s portfolio, enabling products to be placed on trays that move under controlled lighting conditions for effective imaging.

Types of Scanners Available

• The catalog features various scanners designed for different product sizes; some are compact desktop models while others are larger, autonomous systems suitable for high-volume production environments.

Applications in Art and Defense

Scanning Artwork

• Specialized scanners can be mounted vertically to scan paintings, allowing users to obtain hyperspectral data through systematic sweeps across the artwork.

AISA Product Line for Embedded Applications

• The AISA line is designed for airborne applications, enabling hyperspectral scanning from aircraft. This technology can analyze crop health or detect camouflage in defense scenarios.

Recent Innovations: AFX Series

Features of the AFX Cameras

• The new AFX series integrates onboard processing capabilities with satellite geolocation systems, making it suitable for drone installations and enhancing operational efficiency.

Lighting Requirements for Hyperspectral Imaging

Importance of Proper Illumination

• Effective illumination is crucial; halogen lights remain recommended indoors due to their broad spectrum coverage. For outdoor aerial applications, sunlight is preferred despite its variability.

Advancements in LED Solutions

• While there are emerging LED solutions covering visible and near-infrared spectra (400 - 1000 nm), they currently lack the power needed for rapid applications compared to traditional halogen lighting.

Limitations of Current Technologies

• Although LEDs have made strides in certain spectral ranges (900 - 1700 nm), they do not yet provide complete spectral coverage necessary for true hyperspectral imaging. Thus, halogen remains the best option today.

Analysis of Spectral Solutions in Laboratory Settings

Recommendations for Laboratory Analysis

• The speaker recommends using halogen lighting for laboratory analyses to obtain a complete spectral spectrum, which provides comprehensive information necessary for decision-making regarding the final solution needed.

• There is a discussion about the high costs associated with infrared milk analysis, which limits complex solutions in current practices.

Equipment and Measurement Techniques

• The speaker suggests that wood particles like chips and bark can be measured simultaneously for caloric content and moisture, indicating potential versatility in measurement techniques.

• A mention of various components essential for hyperspectral systems, including cameras, scanners, illuminations, and software tailored specifically for spectral analysis.

Software Capabilities in Spectral Analysis

• Conventional analysis software is primarily designed for code reading and 2D/3D measurements; however, specialized software is required to analyze vertical lines produced by spectral cameras effectively.

• The process involves creating an analytical model from previously captured sample images to convert hyperspectral images into false-color representations that indicate different materials within a sample.

Real-Time Analysis Innovations

• New market solutions allow real-time analysis rather than just laboratory use. This innovation enables users without extensive experience in hyperspectral analysis to configure systems easily.

• An example from an Austrian manufacturer named Perception Park illustrates how their tool facilitates both laboratory work and online applications.

Working with Backlight Techniques

• It is confirmed that backlighting can be used effectively; this method requires careful setup to ensure light passes through the material being analyzed.

• Emphasis on the necessity of halogen lighting when conducting hyperspectral analyses to ensure accurate results as light must penetrate the sample adequately.

Challenges in Hyperspectral Analysis Setup

• The speaker discusses challenges related to reflection methods versus transmission methods during spectral analysis setups, highlighting issues with slow processing speeds due to equipment configurations.

• A practical example involving coffee beans mixed with stones demonstrates heterogeneity within samples, showcasing variability that can affect analytical outcomes.

Analysis of Contaminants in Coffee Processing

Differentiating Contaminants from Coffee

• The discussion highlights the various types of stones found in coffee processing, emphasizing the need to differentiate contaminants like stones and wood from actual coffee beans. This distinction is crucial for quality control in coffee production.

Real-Time Analysis System

• A real-time analysis system was demonstrated, which uses color differentiation to identify coffee (marked in red), stones (in blue), and wood pieces (in green). This method addresses the variability of coffee products effectively.

Integration with Conventional Vision Systems

• The analyzed images can be sent to conventional vision systems that identify contaminant positions by recognizing specific pixel colors associated with each type of material, enhancing sorting efficiency.

Portable Hyperspectral Camera Solution

• Introduction of a portable hyperspectral camera that operates autonomously within a wavelength range of 400 to 1000 nanometers, allowing for field use without external equipment or power sources. This innovation simplifies data collection in various environments.

Autonomous Scanning Capabilities

• The camera integrates scanning technology and storage capabilities internally, enabling automatic hyperspectral image generation without needing external movement mechanisms. Users can set it up easily on a tripod for immediate operation.

Advanced Applications and Future Prospects

Intelligent Imaging for Plant Analysis

• The camera's software allows users to analyze captured images for specific applications such as detecting fungi or chlorophyll levels in plants, showcasing its versatility beyond just food safety applications.

Autonomy and Usability Features

• Emphasizing the autonomy provided by built-in batteries and internal systems, this device eliminates the need for laptops or additional scanners during fieldwork while still requiring proper lighting conditions when used indoors.

Challenges in Defect Detection

Limitations of Hyperspectral Imaging

• Addressing inquiries about detecting defects in thermosealing processes reveals challenges; specifically, identifying air gaps between layers may require alternative techniques since air lacks a spectral signature detectable by hyperspectral imaging methods.

Alternative Techniques Suggested

• Thermography is mentioned as a potential solution for identifying areas where seals have not formed correctly due to temperature differences, although its viability depends on specific application contexts.

Discussion on Lighting Solutions

Environmental Concerns and Alternatives

• The speaker expresses skepticism about using image and pre-spectral solutions as a viable answer to current lighting issues, indicating that while there are ongoing efforts to eliminate halogen lighting due to environmental regulations, complete solutions are not yet available.

• There is acknowledgment of the challenges posed by halogen lighting in terms of environmental compliance, but it is noted that alternatives are being sought within the market, particularly for specific applications.

Infrared Technology Insights

• The discussion shifts towards medium wave infrared technology, highlighting that beyond 2500 nanometers, traditional light sources like bulbs are replaced with heat-emitting devices known as qualifiers.

• The speaker describes these qualifiers as functioning similarly to heating elements rather than conventional lights, emphasizing their unique operational characteristics at wavelengths exceeding 3000 nanometers.