Articles tagged with Image Processing
Researchers create physical objects from imaging datasets using a new data processing method that preserves fine details and allows quick distinction between parts. This innovation aims to make 3D printing more accessible and allow anyone to print nearly anything.
MIT engineers create a new virtual-reality training system for drones, enabling vehicles to navigate through an empty physical space while 'seeing' a rich, virtual environment. The system, dubbed Flight Goggles, reduces the number of crashes that drones experience in actual training sessions.
Recent studies by Case Western Reserve University's Anant Madabhushi show that his diagnostic imaging lab's 'deep learning' computers can accurately diagnose heart failure and detect various cancers. The machines offer valuable tools for pathologists and radiologists, helping them become more efficient in their work.
Researchers at Massachusetts General Hospital developed an artificial intelligence technique, AUTOMAP, that enables the production of high-quality images in less time and with lower doses. This approach uses deep learning to automatically determine the correct image reconstruction algorithm, allowing for instant feedback during scanning.
The Stanford team has developed an efficient algorithm to process final images from non-line-of-sight imaging, overcoming a significant challenge in capturing 3-D structure of hidden objects. The system can produce images of out-of-view objects in under a second and is computationally efficient enough to run on regular laptops.
Scientists at the University of Washington created a hybrid optics system combining ultrathin metalenses with computational processing to produce high-efficiency full-color images. This innovation overcomes the narrow wavelength range limitation of traditional metalenses, enabling improved image quality and resolution in various applic...
The Tomato Expression Atlas provides a comprehensive map of gene expression across all tissues and developmental stages of the tomato fruit. This database offers unprecedented spatiotemporal resolution and interactivity, enabling researchers to investigate biological processes important for fruit development.
Researchers developed a compact and inexpensive camera that produces high-resolution 3D images from a single 2D image. The DiffuserCam uses computational imaging to reconstruct 100 million voxels from a 1.3-megapixel image, with potential applications in brain research, self-driving cars, and machine learning.
Research at Kumamoto University found that MCI patients exhibit weakened short-term face memorization ability and distinct gaze behavior when trying to recall faces. This study may lead to early detection of dementia, particularly Alzheimer's disease, which is characterized by cognitive decline in daily life.
Scientists at the Marine Biological Laboratory developed a technique using mirrored cover slips to improve the speed and efficiency of light-sheet microscopy. The method doubles the speed of the microscope and markedly improves its efficiency, useful for imaging fast-moving biological processes.
A recent study by the University of Bonn investigates how some signals dissipate along the processing path to conscious perception. The researchers found that the distinction between conscious and unconscious processing follows significantly further down the processing stream than many researchers have been suspecting.
Researchers at MIT's CSAIL developed a new system called CodeCarbonCopy that allows programmers to transplant code from one program into another. The system can translate between different data representations used by the donor and recipient programs, making it possible to import code seamlessly.
Researchers are developing an autonomous detection system for divers and mines to secure underwater infrastructure. The system combines active acoustic remote detection with target recognition by an approaching Autonomous Underwater Vehicle (AUV), improving coverage and response times.
Researchers from MIT and Google developed a machine-learning system that can retouch images in real-time, producing visually indistinguishable results from existing algorithms. The system processes low-resolution images, reducing time and energy consumption, and can be applied to a range of styles, including high-dynamic-range images.
A team of UC Berkeley researchers has developed a new technique for real-time user-guided colorization using deep neural networks. The system enables novice users to quickly produce reasonable results by learning common colors for different objects and making recommendations in real-time.
ORNL's software significantly speeds up microscopy image analysis, allowing researchers to explore how neurons migrate in the brain. The lab also made bioenergy advancements by identifying a bottleneck in breaking down woody plants for biofuels, and developed an ecosystem modeling method that can pinpoint key uncertainties.
Recent advances in quantum image processing (QIP) have improved computing speed, guaranteed security, and minimal storage requirements. QIP technologies utilize entanglement and parallelism to capture, manipulate, and recover quantum images.
Researchers at TU Wien have successfully switched individual oxygen molecules between a reactive and unreactive state using a force microscope. This process enables new possibilities for investigating the inner workings of photocatalysts.
The robber fly has exceptional vision thanks to a unique arrangement of lenses, enabling it to spot prey over half a metre away in under half a second. Its 'lock-on' strategy allows it to adapt mid-air and catch its prey with precision, making it a potential model for drone design.
Researchers found that self-guided emotional imagery training can improve emotional wellbeing in healthy individuals by reducing depressive symptoms and increasing satisfaction with life. The technique was also associated with changes in brain activity, including increased connectivity between image processing networks.
A new automated system, called CRAQ, detects cracks in the steel components of nuclear power plants using an advanced algorithm and machine learning technique. The system outperformed two others under development, providing more robust results by processing multiple video frames and filtering out falsely detected cracks.
The study uses label-free spatial light interference microscopy (SLIM) to image single microtubule dynamics without added dyes or stains. This allows for long-term imaging of cells, enabling the monitoring of protein movement and consumption of ATP.
Researchers at KAUST developed ScaleMine, a system that accelerates frequent subgraph mining (FSM) by up to ten-fold, allowing for faster analysis of large graph data. The new approach uses a two-step process to divide and conquer the search space, resulting in significant performance improvements.
A study published in PNAS used MEG and computational methods to measure real-time brain processes converting face appearance into recognition. The research team mapped brain activity for each point in time, identifying which parts of the brain encode appearance-based information versus identity-based information.
UC Riverside scientists created astrophotography classes for non-science students, resulting in improved understanding of telescopes and cameras, as well as renewed interest in astronomy. The cost-effective courses also encouraged students to take up astrophotography as a hobby, opening the path to future amateur astronomers.
A new 'deep rendering mixture model' enables computers to learn about the visual world largely on their own, using a semisupervised learning method. The algorithm achieved higher accuracy than previous methods in distinguishing handwritten digits, even with limited training examples.
Researchers used magnetoencephalography to decode neuronal activity from MEG recordings, confirming theoretical predictions and showing that invisible images can be partially maintained in high-level brain regions. This study challenges current understanding of conscious perception and demonstrates the power of machine learning tools.
Researchers at MIT's CSAIL develop a sound-recognition system that outperforms predecessors without requiring expensive hand-annotated data. The system is trained on video and achieves high accuracy rates, with applications in improving mobile device context sensitivity and situational awareness of autonomous robots.
A new light-based technique creates secure, invisible watermarks that can be used to detect and prosecute counterfeiting. The technique uses a complex pattern of light as a unique watermark, which is embedded into the content to be protected.
Researchers at Tohoku University found that visual attention has multiple functions and stages, including early visual processing and selective extraction of information. The study proposes a model of spatial attention that can predict different attention effects for various visual processes, which is useful for complex tasks like driv...
A team of researchers from Ocean University of China used logical stochastic resonance to improve the quality of underwater images, enabling better object detection. The approach overcomes challenges in processing degraded images through conventional methods.
A multidisciplinary team has developed a new method to automate the screening of breast cancer histopathology images, improving detection rates. The technique uses image-processing methods to identify cancerous cells, overcoming challenges posed by clustered and vague boundaries.
Physicists from UT Arlington are developing a multifunctional platform to integrate imaging and photo-induced cancer therapy in a single, portable device. The BIGLITE device aims to improve the efficacy and safety of photo-induced therapies by precisely killing cancer cells while sparing healthy ones.
A novel nontoxic process generates larger ultrathin sheets of 2-D nanosheets, increasing the material's surface area by 20 times, which could expand its commercial applications. The controlled gas exfoliation process separates 2D nanomaterials for use in separation and catalysis.
The study reveals alternating periods of high- and low-wavelet entropy (WS) in rat V1 during image processing, indicating dynamic LFPs with synchronized and complexly ordered activity. The parameters RWE and WS quantify neural population activity characteristics that may help decipher visual processing and object recognition.
Researchers at Rice University developed RedEye, a technology that can provide computers with continuous vision, enabling wearables to see and remember what their owners need. By analyzing analog signals in real-time, RedEye improves energy efficiency and presents unique privacy advantages.
Researchers at MIT and University of Georgia have developed an affordable eye-tracking system using machine learning and crowdsourced data. The system achieves a margin of error of 1.5 cm, outperforming previous experimental systems.
The Rochester Digital Cloak uses flat screen displays to extend the range of angles that can be hidden from view. It enables cloaking large objects without requiring expensive lenses.
Researchers developed a platform for autonomous aerial refueling of UAVs, utilizing binocular vision-based sensor and navigation system. The system allows the tanker UAV to capture images of the receiver UAV and estimate its pose, then control the boom towards the receptacle for refueling.
For the first time, researchers have filmed 'living' nuclear pore complexes in action using an ultra-fast atomic force microscope. The study reveals the dynamic behavior of molecular 'tentacles' inside the pore, which regulate the transport of molecules into and out of the cell nucleus.
University of Wisconsin-Madison engineers developed a biologically inspired artificial eye that can see in the dark using a lobster-inspired fish-eye design. The system improves image-taking through lenses rather than sensor components, resulting in fourfold sensitivity improvement.
Researchers have developed Expansion Mini-Microscopy (ExMM), a technique that combines physical expansion with optical magnification to achieve high-resolution images at low costs. This innovation enables the creation of mini-microscopes from webcam and off-the-shelf components, offering a cost-effective solution for imaging biological...
Researchers at NIST and ORNL have developed a new microwave imaging technique that allows for the visualization of processes occurring at boundaries between liquids and solids. This approach enables the study of technologically and medically important processes without damaging samples or interfering with the process being studied.
Researchers develop a novel microfluidic device called the 'IVF chip' that enables high-resolution imaging of the initial steps of fertilization. The device allows scientists to observe the fusion of sperm and egg, membrane remodeling, and sperm DNA incorporation into the egg.
Researchers at NC State University have developed an algorithm that can quickly and accurately reconstruct hyperspectral images using less data. This breakthrough enables faster imaging times and reduced memory requirements, making it suitable for applications such as security, defense, environmental monitoring, and agriculture.
A handheld microscope can capture details up to a half millimeter beneath the tissue surface, where some types of cancerous cells originate. Researchers expect to test it as a cancer-screening tool in clinical settings next year.
A new experimental paradigm aims to investigate the neural mechanism behind human imagination by synchronizing groups of neurons, known as neuronal ensembles. By combining familiar images or concepts, the brain triggers increased firing rates and synchronization of activities in object-specific neurons.
Researchers at Brown University have developed an algorithm that allows off-the-shelf cameras to capture high-quality 3D images without synchronized projector and camera hardware. The technique uses a structured light flash and can assemble images pixel by pixel to create accurate 3D models.
Researchers from Monash University have developed a new image modulation technique called SWIFT to study how the brain categorizes objects. The study found that SWIFT can stimulate specific neural activity patterns in the brain, enabling greater understanding of human visual processing and its potential applications.
Researchers create a new method for generating training images to simulate natural variability, allowing for more accurate predictions and decision-making in various activities such as environmental cleanup and oil drilling. By analyzing a series of images using the demon algorithm, scientists select representative snapshots that captu...
Engineers at MIT have designed an atomic force microscope that scans images 2,000 times faster than existing models, capturing chemical processes taking place at the nanoscale in near-real time. The instrument produces high-resolution 'movies' of condensation, nucleation, dissolution, and deposition of material.
Researchers trained pigeons to distinguish between benign and malignant breast histology and radiology images, showing their ability to generalize what they learned. The study suggests that pigeons could aid in understanding human medical image perception and performance assessment.
Researchers developed a system that sends highly compressed images to servers, which then process the images and send back modified files with reduced bandwidth consumption. The technique reduces bandwidth usage by over 98% and energy consumption by up to 85%, making it ideal for mobile image processing applications.
New York University researchers discovered a neural match for the complexity of visual activity, explaining how brains extract information from images. The study found that diverse neural groups transmit more information when responding to real-world images.
A brain imaging study found that certain facial patches prefer both faces and bodies to process an individual. This suggests the brain combines facial and body information to represent a whole person. The findings have relevance for understanding human social processing.
Researchers discovered that intrinsic neural connections can be used to help the blind detect their environment, making tasks easier and more intuitive. The study used a sensory substitution device to translate images into sound, allowing blind people to associate different sounds with features of their environment.
Researchers have developed a hand-held optical scanner that can image breast tissues in real time, offering benefits over mammography without ionizing radiation. The device uses near-infrared laser diode source to map optical absorption, which may indicate higher blood flow due to abnormalities like tumors.
The new EEFLUX application allows farmers to view field-scale maps of water consumption in near real-time, providing insights into evapotranspiration and water management. With Landsat satellite imagery, experts can track water consumption year over year, making it easier to monitor the effectiveness of conservation projects.
A team of engineers at Duke University has successfully demonstrated a previously unrecognized 3D imaging capability using modern digital cameras by repurposing its existing components. The research uses the image stabilization and focus modules to achieve similar results without additional hardware.
Research published in the International Journal of Biomedical Engineering and Technology reveals a wavelet transform approach that improves medical imaging. The technique boosts the signal-to-noise ratio and reduces artifacts, allowing for earlier disease detection and diagnosis.