The identification of fish species and their physical and chemical characterization play a crucial role in the fishing industry, fish-food research and the management of marine resources. Traditional methods for species identification, such as expert observation, DNA barcoding and meta-barcoding, though effective, require labor-intensive laboratory work. Consequently, there is a pressing need for more objective and efficient methodologies for accurate fish species identification and characterization. This study proposes the use of multivariate analysis and visible-near infrared hyperspectral imaging (HSI) for a rapid characterization of fish, including the evaluation of specific morphological regions of interest (ROIs) in fish images or intrasample spectral variability, species differentiation, and freshness assessment. The study involves three pelagic species: sardine (Strangomera bentincki), silverside (Odontesthes regia) and anchovy (Engraulis ringens). Principal component analysis (PCA), support vector machine regression (SVM-R), partial least squares regression (PLS-R), and partial least squares discriminant analysis (PLS-DA) were applied as multivariate techniques for these purposes. Comparative studies of morphological ROIs revealed significant differences between the spectral characteristics of various fish zones. A decrease in reflectance intensity due to freshness loss was detected, and the prediction of this freshness, quantified as “time after capture,” was achievable using SVM-R, with a 9% relative error of prediction. Overall, VIS-NIR HSI, supported by multivariate analysis, enables differentiation between the studied species, highlighting its potential as a robust fish species identification and characterization tool.

In this study, we designed an automated classification method, inspired by human taxonomic principles, to distinguish visually similar species of pelagic fish in images through the integration of morphological feature analysis with a hierarchical classification technique. By adapting the Keypoint R-CNN model for automated extraction of morphological characteristics, we accurately classified images of anchovies, mackerel, jack mackerel, and sardines, outperforming the results of the direct use of deep learning-based computer vision algorithms. Our method includes taxonomic analysis, exploiting geometric characteristics such as distances and angles between key body parts, segmenting patterned areas, and extracting texture features. Furthermore, we developed hierarchical classification models that employ a dichotomous key based on these key morphological traits to assess specific fish features such as size, shape, mouth orientation, and color patterns, simulating taxonomic classification. We achieved macro-precisions of up to 1.00 for small fish species and 0.98 for larger species, highlighting the pivotal role of keypoint detection combined with hierarchical classification in addressing challenging taxonomic tasks in marine organisms, and providing a scalable and adaptable solution for further applications.

Pelagic fish have evolved specialized biogenic multilayer reflectors composed of stacks of intracellular anhydrous guanine crystals separated by cytoplasm, giving notorious silvery appearance to their skin. While the reflective properties of guanine crystals and their utility for fish camouflage have been shown in other fish species, this is the first evaluation on fish species from the southern hemisphere, and from the Humboldt current system. This is one of the most productive systems on earth, having particular oceanographic conditions such as upwelling, and thus under strong selection pressures. In this study, we conducted a comparative analysis of four pelagic species, Sardine, Anchovy, and Snoek, known for their silvery characteristics, and Mote sculpin, which lacks silvery features. We aimed to explore the biological mechanisms underlying light reflectivity in fish species and to understand how fish skin microstructures affect whole fish light reflectance and intensity in the visible spectrum. We measured the reflectance of individual fish using hyperspectral imaging and characterized the guanine crystal/cytoplasm layers within the skin of each fish using high-resolution scanning electron microscopy. These Scanning Electron Microscopy (SEM) images were analyzed using the 2D discrete Fourier transform to extract the spatial patterns that govern the light interaction with the guanine crystal structures. A novel spatial frequency analysis approach applied to SEM images explained reflectance differences between species with similar spectral behavior. Furthermore, this study presents the first fish classifiers based on the analysis of spatial frequency features, achieving up to 92.14% accuracy using a K-Nearest Neighbors classifier, highlighting the functional and taxonomic relevance of guanine microstructure organization. Our findings confirm, on pelagic fish species from the Humboldt current system, that silvery species have a chaotic distribution/arrangement of guanine crystals, whereas non-silvery species have a more organized arrangement. Accordingly, Fourier analysis indicated that silvery fish are capable of scattering light uniformly across the visible spectrum. In contrast, the Mote sculpin shows a stronger scattering of red light, distinguishing it from silvery fish.

A novel method, to our knowledge, for monitoring weightlifting exercises based on infrared imaging is proposed in this work. For the infrared workout weightlifting recorded scenes, radiometry and artificial intelligence were employed for in-scene temperature and biomechanical athletes' body parts position mapping. Our method was effective in monitoring muscle exertion during high-performance athletic exercises, as evidenced by the results obtained from real athletic datasets. The method generates a color-labeled sequence of thermal images and reports on body part positions, which can be used by judges and trainers to guide athletes toward safer and more efficient practices. (c) 2024 Optica Publishing Group. All rights, including for text and data mining (TDM), Artificial Intelligence (AI) training, and similar technologies, are reserved.

In this work, we will share the main results achieved with a Long Wave Infrared (LWIR) Light Field (LF) imaging system with two novel capabilities relevant to IR image science applications: The capability of digitally refocusing to any nearby object planes with a high Signal to Noise Rat io (SNR), this is, to achieve refocused image object planes almost free of Fixed-Pattern Noise (FPN) and blur artifacts. And, the capability of achieving multispectral LWIR imaging for the global scene and for all the refocused nearby object planes required, this is, LWIR radiometry refocusing capacity. The built-in LWIR LF imaging system is implemented with an LWIR microbolometer Xenics camera 8-12 micrometers spectral band, and a high precision scanning system(Newport). LWIR multispectral capacity is achieved with an array of narrow-band LWIR interference optical filters.

We propose a single-camera multispectral Short Wave Infrared (SWIR) plenoptic imaging system free of fixed pattern noise (FPN). The developed system can form, for each scene’s object of interest, an in-focus noiseless 24 channels SWIR multispectral data cube. The Fourier Slice Photography Theorem is employed in the integrated multispectral light field to achieve images at a specific scene depth of focus. And, a reference-free image quality index is used for selecting the in-focus image for each scene object focal plane. Results are then compared to SWIR multispectral data cubes obtained from light fields achieved after applying to the SWIR camera a blackbody two-point calibration method. The scientific potential of the imaging system is proved by imaging thermal stressed plants at different scene depths of focus. The FPN SWIR spectral dependence is evaluated on the images digitally refocused at the farthest scene focal plane, resulting in concordance with the InGaAs FPA quantum efficiency. The lightfield (LF) system performance to naturally filter out the FPN was evaluated with a Black body (BB) reference base index, resulting in an SSIM index close to 98 % at the object image focus plane.

Fishing landings in Chile are inspected to control fisheries that are subject to catch quotas. The control process is not easy since the volumes extracted are large and the numbers of landings and artisan shipowners are high. Moreover, the number of inspectors is limited, and a non-automated method is utilized that normally requires months of training. In this work, we propose, design, and implement an automated fish landing control system. The system consists of a custom gate with a camera array and controlled illumination that performs automatic video acquisition once the fish landing starts. The imagery is sent to the cloud in real time and processed by a custom-designed detection algorithm based on deep convolutional networks. The detection algorithm identifies and classifies different pelagic species in real time, and it has been tuned to identify the specific species found in landings of two fishing industries in the Biobío region in Chile. A web-based industrial software was also developed to display a list of fish detections, record relevant statistical summaries, and create landing reports in a user interface. All the records are stored in the cloud for future analyses and possible Chilean government audits. The system can automatically, remotely, and continuously identify and classify the following species: anchovy, jack mackerel, jumbo squid, mackerel, sardine, and snoek, considerably outperforming the current manual procedure.

A single camera infrared plenoptic system free of fixed pattern noise is im-plemented using the roughness Laplacian pattern index. Results are then compared to the two-point calibration method using the structural similarity index.

Datasets of VIS-NIR (201 bands, 565-805 nm) and SWIR (101 bands, 1066-1386 nm) average spectra of 82 pellets made of copper concentrate mixtures. The dataset are splited into 75 % training and 25 % testing. The copper concentrate mixture used in this dataset have been characterized via QEMSCAN in: F. rivas, et al, "QEMSCAN REPORT OF 82 COPPER CONCENTRATES," figshare (2023) https://opticapublishing.figshare.com/s/52c54f5634ad9fe79878 Classification models available for this dataset can be found in F. rivas, et al, "VISNIR SWIR machine learning models for classification of copper concentrates," figshare (2023) https://opticapublishing.figshare.com/s/0e9c395a492f7ec1abde

Real-time temperature surveillance of the reactions and phase transformations in the flash smelting furnaces burner flame is of vital importance to assess the operational process. For this purpose, a radiometric optical system based on a visible to near-infrared (VIS-NIR) spectrometer fit with a specialized method is proposed as a sensor for industrial flash copper smelters, thus providing real-time information to aid process control. The proposed sensor captures the burners flame irradiance to estimate temperature and emissivity using an optimization-based multiwavelength estimation method rooted around Planck’s radiation model. This multiwavelength method (MWM) is capable of calculating radiometric temperature and spectral emissivity, without previous knowledge of the emissivity model. In this work, different optimization algorithms were used to solve the multiwavelength model, and the results are compared with the commonly used two-wavelength pyrometric method for data obtained in laboratory and industrial copper smelter scenarios. The method’s robustness in the presence of additive white noise is studied between −30 and 30 dB signal-to-noise ratio (SNR). The MWM reported minimum temperature error values of 4 °C and a relative error under 10% for an SNR of −30 dB, giving this method essential characteristics for industrial applications, measurement accuracy, and robustness.

Fishing has provided mankind with a protein-rich source of food and labor, allowing for the development of an important industry, which has led to the overexploitation of most targeted fish species. The sustainable management of these natural resources requires effective control of fish landings and, therefore, an accurate calculation of fishing quotas. This work proposes a deep learning-based spatial-spectral method to classify five pelagic species of interest for the Chilean fishing industry, including the targeted Engraulis ringens, Merluccius gayi, and Strangomera bentincki and non-targeted Normanichthtys crockeri and Stromateus stellatus fish species. This proof-of-concept method is composed of two channels of a convolutional neural network (CNN) architecture that processes the Red–Green–Blue (RGB) images and the visible and near-infrared (VIS-NIR) reflectance spectra of each species. The classification results of the CNN model achieved over 94% in all performance metrics, outperforming other state-of-the-art techniques. These results support the potential use of the proposed method to automatically monitor fish landings and, therefore, ensure compliance with the established fishing quotas.

The areas in which the Infrared (IR) spectrum gives vital information are as different as they are unique. However, this versatility comes with limitations. One of these obstacles is how to deal with the noise that affects this spectrum, which heavily distorts the data acquired from a scene. This investigation centers itself around how to fix the Fixed Pattern Noise (FPN) in the Short Wave Infrared (SWIR) using a refocusing algorithm, as an alternative of the two point black body calibration. Plants are chosen because of the ease to identify hidric stress in them using IR imaging due to the water natural reflectance and absorbance properties. The elimination of noise using this method will be revised, comparing it to the well known two point calibration method. Furthermore, by using filters a multispectral cube will be created and a spectral range of interest within the SWIR will be defined.

A study on the classification of copper concentrates relevant to the copper refining industry is performed by means of reflectance hyperspectral images in the visible and near infrared (VIS-NIR) bands (400-1000 nm) and in the short-wave infrared (SWIR) (900-1700 nm) band. A total of 82 copper concentrate samples were press compacted into 13-mm-diameter pellets, and their mineralogical composition was characterized via quantitative evaluation of minerals and scanning electron microscopy. The most representative minerals contained in these pellets are bornite, chalcopyrite, covelline, enargite, and pyrite. Three databases (VIS-NIR, SWIR, and VIS-NIR-SWIR) containing a collection of average reflectance spectra computed from 9×9p i x e l neighborhoods in each pellet hyperspectral image are compiled to train the classification models. The classification models tested in this work are a linear discriminant classifier and two non-linear classifiers, a quadratic discriminant classifier, and a fine K-nearest neighbor classifier (FKNNC). The results obtained show that the joint use of VIS-NIR and SWIR bands allows for the accurate classification of similar copper concentrates that contain only minor differences in their mineralogical composition. Specifically, among the three tested classification models, the FKNNC performs the best in terms of overall classification accuracy, achieving 93.4% accuracy in the test set when only VIS-NIR data are used to construct the classification model, up to 80.5% using only SWIR data, and up to 97.6% using both VIS-NIR and SWIR bands together.

This research shows a prototype for crowd location and counting for earthquakes based on deep learning and the infrastructure of a state-of-the-art 5G standalone network deployed at the Universidad de Concepcion, Chile. The system uses an 8 MP panoramic network camera to capture real-time crowd images, which are sent to a Deep Learning Server (DLS) over the 5G network. The camera provides visible color images, and its sensor technology can provide color images even at night. The DLS uses frames from the video feed and generates Focal Inverse Distance Transform (FIDT) maps, in which the counting and location of people are carried out. In particular, the FIDT maps are generated from the crowd images using a deep-learning model composed of two cascaded autoencoders. The 5G technology allows the system to transfer data from the camera to DLS at high speed, an essential feature for a system that will help authorities make critical decisions during natural disasters. Under this scenario, and considering that the number of rescuers is usually limited, our system enables a better distribution of them among several crowded places by instantly knowing the number of people at any time of the day or night.

This paper presents SAFE, a prototype system for supporting the fish landings control of small-scale fishing boats in Chile. SAFE is a modern solution for fishery inspection that automatically discriminates fish species using machine learning. Here, we present a version of SAFE that classifies five target pelagic fish species in Chile: anchovy, Chilean jack mackerel, hake, mote sculpin, and sardine. The system has two stages; the first detects and segments all fish appearing in an image. These segmented images then feed the second stage, which perform species classification. A database of approximately 266 images from these five fish species was constructed for training, validation, and testing purposes. For the fish detection stage, we exploited transfer learning to train Mask R-CNN architectures, an instance segmentation model. As for the fish species classification stage, we exploited transfer learning to train ResNet50 and VGG16 deep learning architectures. Results show that SAFE achieves between 90% and 96.3% macro-average precision (MP) when classifying the five fish species mentioned above. The best architecture, composed of a Mask R-CNN-based detector and a VGG16-based classifier, achieves an MP of 96.3%, which could process a single fish as quick as 16.67 FPS, and one whole 1920x1080-pixel image as quick as 2 FPS.

This paper proposes crowd estimation technology to help authorities make the right decisions in times of crisis. Specifically, deep learning models have faced these challenges, achieving excellent results. In particular, the trend of using single-column Fully Convolutional Networks (FCNs) has increased in recent years. A typical architecture that meets these characteristics is the autoencoder. However, this model presents an intrinsic difficulty: the search for the optimal dimensionality of the latent space. In order to alleviate such difficulty, we propose a dual architecture consisting of two cascaded autoencoders. The first autoencoder is responsible for carrying out the masked reconstruction of the original images, whereas the second obtains crowd maps from the outputs of the first one. In this way, our architecture improves the location of people and crowds in Focal Inverse Distance Transform (FIDT) maps, resulting in more accurate count estimates than estimates obtained through a single autoencoder architecture.

A three-dimensional point spread function experimental estimation method based on the system’s focal plane array spatial local impulse response of a mid-wave infrared microscope is presented. The method uses several out-of-focus two-dimensional point spread function planes to achieve a single three-dimensional point spread function of the whole microscope’s optical spreading, expanding the limits of infrared optical technology by one dimension. This technique includes stages of image acquisition, nonuniformity correction, filtering, and multi-planar reconstruction steps, and its effectiveness is demonstrated on biological sample image restoration by means of a multi-planar refocusing application.

This article presents an algorithm to improve spatial resolution from a sequence of multispectral microscopy images cubes. We propose to use spatial regularization terms that exploit the spatial-spectral structural similarity of the observed image.

Earthquakes, and their cascading threats to economic and social sustainability, are a common problem between China and Chile. In such emergencies, automatic image recognition systems have become critical tools for preventing and reducing civilian casualties. Human crowd detection and estimation are fundamental for automatic recognition under life-threatening natural disasters. However, detecting and estimating crowds in scenes is non-trivial due to occlusion, complex behaviors, posture changes, and camera angles, among other issues. This paper presents the first steps i n developing a n intelligent Earthquake Early Warning System (EEWS) between China and Chile. The EEWS exploits the ability of deep learning architectures to properly model different spatial scales of people and the varying degrees of crowd densities. We propose an autoencoder architecture for crowd detection and estimation because it creates compressed representations for the original crowd input images in its latent space. The proposed architecture considers two cascaded autoencoders. The first performs reconstructive masking of the input images, while the second generates Focal Inverse Distance Transform (FIDT) maps. Thus, the cascaded autoencoders improve the ability of the network to locate people and crowds, thereby generating high-quality crowd maps and more reliable count estimates.

Coupling HSI and μ-LIBS for elemental and mineralogical imaging in rocks. Elemental and mineral distribution with micrometric spatial resolution. μ-LIBS was expanded to a new field of molecular imaging.