At OpenScience, we are constantly innovating to offer accessible and high-impact solutions to the challenges in the agricultural and scientific sectors. We have recently developed a focus correction method designed to transform images with varying sharpness levels into a detailed 3D representation. These images are captured using the multispectral microscope also developed by OpenScience.
Multispectral Microscope
The technology behind our multispectral microscope is capable of capturing images across various ranges of the electromagnetic spectrum, spanning from visible regions (the colors we can perceive with the naked eye) to ultraviolet (UV) and near-infrared (NIR). Each of these ranges provides unique information about the sample's properties, such as its chemical composition and surface structure. The microscope uses specific LED light sources to illuminate the sample at different wavelengths. Equipped with an automatic sampler, the system allows continuous analysis of multiple samples. The samples, such as grains or particles, are placed on an analysis platform, and the microscope can be configured to capture images at different positions on the X, Y, and Z axes (different heights).
Image of OpenScience's multispectral microscope
Why is focus correction necessary?
The need to implement a focus correction method in a multispectral microscope is related to the characteristics of the different ranges of the electromagnetic spectrum and the properties of the samples. Below are the main reasons for the development of a focus correction method:
Dispersion and Chromatic Aberration
Different wavelengths of light are refracted in different ways when passing through lenses, causing shifts in the focal plane. For example, an image captured in the near-infrared (NIR) may have a different focal plane than an image obtained in the visible range. Focus correction ensures that all images are sharp and aligned, regardless of wavelength.
Variation in Sample Structure
Samples may have irregular surfaces or different layers of interest (e.g., particles with roughness or grains with varying heights). Focus correction allows adjusting the focal plane to observe specific details of each layer or characteristic of the sample.
Multiplane Configurations
In some applications, it is necessary to capture images at different focal planes for 3D reconstructions or to study the depth of structures. Focus correction is essential to ensure each plane is captured accurately.
Improvement of Data Quality
Properly focusing on each spectral range improves image sharpness and the accuracy of information extraction, such as chemical spectra or structural features. Blurry images can compromise analysis and lead to erroneous interpretations.
Automation and Efficiency
Automatic focus correction methods (such as contrast-based autofocus or laser) reduce the need for manual intervention, speeding up data acquisition and improving reproducibility.
Experiment Conducted to Evaluate the Effectiveness of the Method
For the development of the multispectral microscope's focus correction algorithm, an experiment was conducted using a broached coffee bean as a sample. This choice was strategic, considering that OpenScience already has extensive experience and several works related to the analysis of coffee beans.
Experiment Procedures
Multispectral Image Capture
Images of the grain were captured at different height variations. This approach allowed images to be recorded at multiple focal planes, creating a robust dataset to evaluate the algorithm's effectiveness in identifying the optimal focal plane.
Specific Relevance of the Broached Coffee
In the selected grain, the presence of the borer and internal tunnels represented an ideal condition to test the microscope's and algorithm's ability to detect structural details and variations that directly affect coffee quality. These damages are difficult to identify with the naked eye or in superficial analyses but are critical for the industry as they impact flavor and the price of the final product.
Example of an out-of-focus image of a broca-damaged coffee bean
Experiment Objective
The main objectives of the experiment were:
To capture multispectral images of the grain at different heights to perform 3D and 2D image reconstruction, retrieving information about the grain’s structure, such as texture, roughness, height, and other physical characteristics.
To evaluate whether it is possible to obtain a fully focused image by combining the information from images captured at different heights.
Focus Correction Method
After the experiment, the images were divided into several parts, and the focus quality for each was assessed. Based on this data, a three-dimensional representation was created, reflecting the areas of best focus for each analyzed region.
Additionally, the process preserves the spectral information of the images, ensuring that important details can still be extracted. A crucial step for the image reconstruction was ensuring that all spectral layers were aligned correctly. This alignment was done using methods to ensure all information was synchronized, allowing the images to be reconstructed with optimized focus. The final result is a three-dimensional representation of the best focal heights for each part of the sample, providing a more detailed and accurate analysis.
Reconstructed focus image of a broca-damaged coffee bean
Results and Impact
The developed algorithm offers several advantages, making it a robust and versatile solution for multispectral applications. In terms of accuracy and reliability, it is capable of automatically identifying the optimal heights for each region of the sample, allowing the reconstruction of images with optimized focus, even in samples with varying heights. Additionally, spectral preservation ensures that the reconstructions maintain the integrity of information across all analyzed wavelengths.
In terms of applicability, the algorithm stands out particularly in agricultural and scientific analyses, such as the evaluation of grains and food quality. It also enables the reconstruction of focused multispectral data cubes, essential for chemical and quality analyses, while preserving the spectral information necessary for these purposes with high fidelity.
Another key point is the system's automation and precision, which minimize the need for human intervention, increasing efficiency and reproducibility of the analyses. Furthermore, this method can be applied to a wide variety of samples, such as grains generally damaged by insects, allowing the identification of structural and qualitative defects.
Future Perspectives
The next steps of the project include the full integration of the algorithm into OpenScience's multispectral equipment, enabling automation of the entire process of image capture and focused reconstruction. This integration aims to significantly increase analytical efficiency, reduce the need for manual intervention, and expand the range of applications, especially in the agricultural and scientific sectors.
With this technology, OpenScience's equipment will be ready to provide robust and automated multispectral solutions, optimizing the quality of the data obtained and accelerating analysis processes. This approach not only enables advances in grain and food quality assessment but also paves the way for new applications in other areas, such as industrial inspection, environmental monitoring, and scientific research.
We are committed to collaborating with the scientific and industrial communities to maximize the impact of this innovation. For more information or partnerships, we are available to discuss the advancements of this technology and its transformative potential in multispectral analysis.
Comentários