Underwater spectral imaging is a promising method for mapping, classification and health monitoring of coral reefs and seafloor inhabitants. However, the spectrum of light is distorted during the underwater imaging process due to wavelength-dependent attenuation by the water. This paper presents a model-based method that accurately restores brightness of underwater spectral images captured with narrowband filters. A model is built for narrowband underwater spectral imaging. The model structure is derived from physical principles, representing the absorption, scattering and refraction by water and the optical properties of narrowband filters, lenses and image sensors. The model coefficients are calibrated based on spectral images captured underwater and in air. With the imaging model available, energy loss due to water attenuation is restored for images captured at different underwater distances. An experimental setup is built and experiments are carried out to verify the proposed method. Underwater images captured within an underwater distance of 260 cm are restored and compared with those in air. Results show that the relative restoration error is 3.58% on average for the test images, thus proving the accuracy of the proposed method.
Among aquatic biota, corals provide shelter with sufficient nutrition to a wide variety of underwater life. However, a severe decline in the coral resources can be noted in the last decades due to global environmental changes causing marine pollution. Hence, it is of paramount importance to develop and deploy swift coral monitoring system to alleviate the destruction of corals. Performing semantic segmentation on underwater images is one of the most efficient methods for automatic investigation of corals. Firstly, to design a coral investigation system, RGB and spectral images of various types of corals in natural and artificial aquatic sites are collected. Based on single-channel images, a convolutional neural network (CNN) model, named DeeperLabC, is employed for the semantic segmentation of corals, which is a concise and modified deeperlab model with encoder-decoder architecture. Using ResNet34 as a skeleton network, the proposed model extracts coral features in the images and performs semantic segmentation. DeeperLabC achieved state-of-the-art coral segmentation with an overall mean intersection over union (IoU) value of 93.90%, and maximum F1-score of 97.10% which surpassed other existing benchmark neural networks for semantic segmentation. The class activation map (CAM) module also proved the excellent performance of the DeeperLabC model in binary classification among coral and non-coral bodies.
In the past decade, underwater spectral imaging (USI) has shown great potential in underwater exploration for its high spectral and spatial resolution. This proposal presents a stare-type USI system combined with the liquid crystal tunable filter (LCTF) spectral splitting device. Considering the working features of LCTF and the theoretical model of USI, the core structure containing “imaging lens-LCTF-imaging sensor” is designed and developed. The system is compact, and the optical geometry is constructed minimally. The spectral calibration test analysis proved that the spectral response range of the system covers a full band of 400 nm to 700 nm with the highest spectral resolution between 6.7 nm and 18.5 nm. The experiments show that the system can quickly collect high-quality spectral image data by switching between different spectral bands arbitrarily. The designed prototype provides a feasible and reliable spectral imaging solution for in situ underwater targets observation with high spectrum collecting efficiency.
2 March 2021 in Journal of Marine Science and Application
Monitoring the ecology and physiology of corals, sediments, planktons, and microplastic at a suitable spatial resolution is of great importance in oceanic scientific research. To meet this requirement, an underwater microscope with an electrically controlled variable lens was designed and tested. The captured microscopic images of corals, sediments, planktons, and microplastic revealed their physical, biological, and morphological characteristics. Further studies of the images also revealed the growth, degradation, and bleaching patterns of corals; the presence of plankton communities; and the types of microplastics. The imaging performance is majorly influenced by the choice of lenses, camera selection, and lighting method. Image dehazing, global saturation masks, and image histograms were used to extract the image features. Fundamental experimental proof was obtained with micro-scale images of corals, sediments, planktons, and microplastic at different magnifications. The designed underwater microscope can provide relevant new insights into the observation and detection of the future conditions of aquatic ecosystems.
Vital transportation of hazardous and noxious substances (HNSs) by sea occasionally suffers spill incidents causing perilous mutilations to off-shore and on-shore ecology. Consequently, it is essential to monitor the spilled HNSs rapidly and mitigate the damages in time. Focusing on on-site and early processing, this paper explores the potential of deep learning and single-spectrum ultraviolet imaging (UV) for detecting HNSs spills. Images of three floating HNSs, including benzene, xylene, and palm oil, captured in different natural and artificial aquatic sites were collected. The image dataset involved UV (at 365 nm) and RGB images for training and comparative analysis of the detection system. The You Only Look Once (YOLOv3) deep learning model is modified to balance the higher accuracy and swift detection. With the MobileNetv2 backbone architecture and generalized intersection over union (GIoU) loss function, the model achieved mean IoU values of 86.57% for UV and 82.43% for RGB images. The model yielded a mean average precision (mAP) of 86.89% and 72.40% for UV and RGB images, respectively. The average speed of 57 frames per second (fps) and average detection time of 0.0119 s per image validated the swift performance of the proposed model. The modified deep learning model combined with UV imaging is considered computationally cost-effective resulting in precise detection accuracy and significantly faster detection speed.
Measurement biases arising from changes in temperature can be a major concern when using miniature spectrometers in extreme environments, particularly when temperature stabilization approaches are not feasible. Here, temperature-related biases of a low-power field spectrometry system comprised of a CMOS miniature monolithic spectrometer module and custom driver electronics were examined between and , well below the stated operating range of this particular spectrometer. Using these observations, a predictive model was developed to estimate the dark output of the spectrometry system within this extended operating range. This information was used to correct the signal at any measured integration time and temperature to that which would be measured at a reference integration time and temperature. This approach provides a general framework for assessing the temperature dependence of monolithic spectrometers whose field use will occur at temperatures outside of the range examined by the manufacturer.