More Like this

1. Compressive Sensing-Based Reconstruction of Lissajous-Like Nodding Lidar Data

   https://doi.org/10.1115/DSCC2019-9201  

   Benson, Michael T.; Sathishchandra, Harish; Clayton, Garrett M.; Andersson, Sean B. (October 2019, 2019 ASME Dynamic Systems and Control Conference)

   In this article, a compressive sensing (CS) reconstruction algorithm is applied to data acquired from a nodding multi-beam Lidar system following a Lissajous-like trajectory. Multi-beam Lidar systems provide 3D depth information of the environment for applications in robotics, but the vertical resolution of these devices may be insufficient to identify objects, especially when the object is small and/or far from the robot. In order to overcome this issue, the Lidar can be nodded in order to obtain higher vertical resolution with the side-effect of increased scan time, especially when raster scan patterns are used. Such systems, especially when combined with nodding, also yield large volumes of data which may be difficult to store and mange on resource constrained systems. Using Lissajous-like nodding trajectories allows for the trade-off between scan time and horizontal and vertical resolutions through the choice of scan parameters. These patterns also naturally sub-sample the imaged area and the data can be further reduced by simply not collecting each data point along the trajectory. The final depth image must then be reconstructed from the sub-sampled data. In this article, a CS reconstruction algorithm is applied to data collected during a fast and therefore low-resolution Lissajous-like scan. Experiments and simulations show the feasibility of this method and compare its results to images produced from simple nearest-neighbor interpolation. 

   more » « less
2. Automated plankton classification from holographic imagery with deep convolutional neural networks

   https://doi.org/10.1002/lom3.10402  

   Guo, Buyu; Nyman, Lisa; Nayak, Aditya R.; Milmore, David; McFarland, Malcolm; Twardowski, Michael S.; Sullivan, James M.; Yu, Jia; Hong, Jiarong (December 2020, Limnology and Oceanography: Methods)

   Abstract In situ digital inline holography is a technique which can be used to acquire high‐resolution imagery of plankton and examine their spatial and temporal distributions within the water column in a nonintrusive manner. However, for effective expert identification of an organism from digital holographic imagery, it is necessary to apply a computationally expensive numerical reconstruction algorithm. This lengthy process inhibits real‐time monitoring of plankton distributions. Deep learning methods, such as convolutional neural networks, applied to interference patterns of different organisms from minimally processed holograms can eliminate the need for reconstruction and accomplish real‐time computation. In this article, we integrate deep learning methods with digital inline holography to create a rapid and accurate plankton classification network for 10 classes of organisms that are commonly seen in our data sets. We describe the procedure from preprocessing to classification. Our network achieves 93.8% accuracy when applied to a manually classified testing data set. Upon further application of a probability filter to eliminate false classification, the average precision and recall are 96.8% and 95.0%, respectively. Furthermore, the network was applied to 7500 in situ holograms collected at East Sound in Washington during a vertical profile to characterize depth distribution of the local diatoms. The results are in agreement with simultaneously recorded independent chlorophyll concentration depth profiles. This lightweight network exemplifies its capability for real‐time, high‐accuracy plankton classification and it has the potential to be deployed on imaging instruments for long‐term in situ plankton monitoring. 

   more » « less
3. Design of Multichannel Spectrum Intelligence Systems Using Approximate Discrete Fourier Transform Algorithm for Antenna Array-Based Spectrum Perception Applications

   https://doi.org/10.3390/a17080338  

   Madanayake, Arjuna; Lawrance, Keththura; Kumarasiri, Bopage Umesha; Sivasankar, Sivakumar; Gunaratne, Thushara; Edussooriya, Chamira_U S; Cintra, Renato J (August 2024, Algorithms)

   The radio spectrum is a scarce and extremely valuable resource that demands careful real-time monitoring and dynamic resource allocation. Dynamic spectrum access (DSA) is a new paradigm for managing the radio spectrum, which requires AI/ML-driven algorithms for optimum performance under rapidly changing channel conditions and possible cyber-attacks in the electromagnetic domain. Fast sensing across multiple directions using array processors, with subsequent AI/ML-based algorithms for the sensing and perception of waveforms that are measured from the environment is critical for providing decision support in DSA. As part of directional and wideband spectrum perception, the ability to finely channelize wideband inputs using efficient Fourier analysis is much needed. However, a fine-grain fast Fourier transform (FFT) across a large number of directions is computationally intensive and leads to a high chip area and power consumption. We address this issue by exploiting the recently proposed approximate discrete Fourier transform (ADFT), which has its own sparse factorization for real-time implementation at a low complexity and power consumption. The ADFT is used to create a wideband multibeam RF digital beamformer and temporal spectrum-based attention unit that monitors 32 discrete directions across 32 sub-bands in real-time using a multiplierless algorithm with low computational complexity. The output of this spectral attention unit is applied as a decision variable to an intelligent receiver that adapts its center frequency and frequency resolution via FFT channelizers that are custom-built for real-time monitoring at high resolution. This two-step process allows the fine-gain FFT to be applied only to directions and bands of interest as determined by the ADFT-based low-complexity 2D spacetime attention unit. The fine-grain FFT provides a spectral signature that can find future use cases in neural network engines for achieving modulation recognition, IoT device identification, and RFI identification. Beamforming and spectral channelization algorithms, a digital computer architecture, and early prototypes using a 32-element fully digital multichannel receiver and field programmable gate array (FPGA)-based high-speed software-defined radio (SDR) are presented. 

   more » « less
4. Cra: A Generic Compression Ratio Adapter for End-To-End Data-Driven Image Compressive Sensing Reconstruction Frameworks

   https://doi.org/10.1109/ICASSP40776.2020.9052998  

   Zhang, Zhikang; Xu, Kai; Ren, Fengbo (May 2020, IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP))

   End-to-end data-driven image compressive sensing reconstruction (EDCSR) frameworks achieve state-of-the-art reconstruction performance in terms of reconstruction speed and accuracy. However, due to their end-to-end nature, existing EDCSR frameworks can not adapt to a variable compression ratio (CR). For applications that desire a variable CR, existing EDCSR frameworks must be trained from scratch at each CR, which is computationally costly and time-consuming. This paper presents a generic compression ratio adapter (CRA) framework that addresses the variable compression ratio (CR) problem for existing EDCSR frameworks with no modification to given reconstruction models nor enormous rounds of training needed. CRA exploits an initial reconstruction network to generate an initial estimate of reconstruction results based on a small portion of the acquired measurements. Subsequently, CRA approximates full measurements for the main reconstruction network by complementing the sensed measurements with resensed initial estimate. Our experiments based on two public image datasets (CIFAR10 and Set5) show that CRA provides an average of 13.02 dB and 5.38 dB PSNR improvement across the CRs from 5 to 30 over a naive zero-padding approach and the AdaptiveNN approach(a prior work), respectively. CRA addresses the fixed-CR limitation of existing EDCSR frameworks and makes them suitable for resource-constrained compressive sensing applications. 

   more » « less
5. Tree Species Classification of Forest Stands Using Multisource Remote Sensing Data

   https://doi.org/10.3390/rs13010144  

   Wan, Haoming; Tang, Yunwei; Jing, Linhai; Li, Hui; Qiu, Fang; Wu, Wenjin (January 2021, Remote Sensing)

   The spatial distribution of forest stands is one of the fundamental properties of forests. Timely and accurately obtained stand distribution can help people better understand, manage, and utilize forests. The development of remote sensing technology has made it possible to map the distribution of tree species in a timely and accurate manner. At present, a large amount of remote sensing data have been accumulated, including high-spatial-resolution images, time-series images, light detection and ranging (LiDAR) data, etc. However, these data have not been fully utilized. To accurately identify the tree species of forest stands, various and complementary data need to be synthesized for classification. A curve matching based method called the fusion of spectral image and point data (FSP) algorithm was developed to fuse high-spatial-resolution images, time-series images, and LiDAR data for forest stand classification. In this method, the multispectral Sentinel-2 image and high-spatial-resolution aerial images were first fused. Then, the fused images were segmented to derive forest stands, which are the basic unit for classification. To extract features from forest stands, the gray histogram of each band was extracted from the aerial images. The average reflectance in each stand was calculated and stacked for the time-series images. The profile curve of forest structure was generated from the LiDAR data. Finally, the features of forest stands were compared with training samples using curve matching methods to derive the tree species. The developed method was tested in a forest farm to classify 11 tree species. The average accuracy of the FSP method for ten performances was between 0.900 and 0.913, and the maximum accuracy was 0.945. The experiments demonstrate that the FSP method is more accurate and stable than traditional machine learning classification methods. 

   more » « less