Volume 5, Number 3 May/June 1997
Moving Forward
|
Technology Opportunity Showcase highlights some unique technologies that NASA has developed and that we believe have strong potential for commercial application. While the descriptions provided here are brief, they should provide enough information to communicate the potential applications of the technology. For more detailed information, contact the person or office listed. Please mention you read about it in Innovation. |
Prototype Internet Access Model
NASA's Langley Research Center, as part of its effort in the High Performance Computing and Communications/Information Infrastructure Technology and Applications program, has developed a prototype Internet access model that allows an entire local area network (LAN) of computers to connect to the Internet using only a standard, analog telephone line. By using a powerful network server to provide numerous functions for the LAN and using an intelligent network-based disk caching scheme, the cost of connecting to the Internet has been reduced by up to 80 percent over standard connectivity solutions for public schools in Virginia's Tidewater region. The model is scalable with up to 253 computers connected to the Internet simultaneously. Also, as demand dictates, the analog phone line can be replaced by a higher speed digital connection.
Lossless DCT (Discrete Cosine Transform)
NASA's Goddard Space Flight Center has developed Lossless DCT (Discrete Cosine Transform)-based Lossy algorithms that can be used for the analysis of medical images. The Lossless algorithm is an extension of the Rice algorithm that allows the compression of low entropy as well as normal and high entropy images. This algorithm is currently under consideration by the Consultative Committee on Space Data Systems to become an international space standard for lossless compression. The Lossy algorithm uses hybrid transform to reduce blocking distortions inherent in two-dimensional DCT. This system employs an adaptive encoder, which reduces the overhead associated with Hoffman code tables.
Registration and Data Fusion Tools
NASA's Jet Propulsion Laboratory has developed registration and data fusion tools designed originally to emerge tomographic, elevation and seismic data sets gathered from satellites and distribution sensors. The laboratory is currently adapting these tools for clinical medicine for fusing volumetric data sets from CT, MRI, PET and SPECT scanners. Pattern recognition algorithms also have been developed that can automatically classify and segment various regions of a fused volumetric medical image as normal or abnormal, based on histology-confirmed training data. This technology has numerous applications in oncology, orthopedic surgery and opthalmology.
ROSS 3D Reconstruction Software
This Ames Research Center technology, originally designed for use with biological cells, tissues or organs, allows for the reconstruction of any complex three-dimensional (3D) object that can be imaged in sections or layers by physical, optical, sound or other methods. Unlike other sectional modeling systems, the ROSS 3D Reconstruction Software acquires data directly from the imaging source (confocal microscopy, electron microscopy, ultrasound, and so on) without the use of photo- or radio-sensitive films. Features of the software include the fastest means of capturing images from a transmission microscope or other imaging probe, automated calibration of microscope stage parameters during image capture, electronic acquisition of complete data sets, remote sharing of data and arbitrary sectioning capable of rendering complex, branched objects.
VSLI Implemented Neural Networks
The Jet Propulsion Laboratory has developed VLSI-implemented neural networks that perform motion estimation and image data compression for medical applications. The system processes video image data, transmitting only the nonredundant parts in an efficient data stream, and consists of a motion estimation processor and an image compression processor implemented by VLSI circuitry. The motion estimation neuroprocessor implements a neuro-network-based motion-estimation algorithm to achieve a high-speed, wide-range estimation of motions in images.
DCT Algorithms
NASA's Ames Research Center has developed DCT algorithms that minimize noise while maximizing compression. This performance is obtained by optimizing the DCT quantization coefficients at the threshold of visibility, taking into account luminance, veiling light, spatial frequency and special frequency-related conditions (pixel size, viewing distance and aspect ratio). This DCT model pools errors nonlinearly over the image to yield perceptual error. The model provides the maximum visual quality for a given bit rate. It also provides the user with a sensible and meaningful quality scale for other DCT-based algorithms.
For more information about these technology opportunities,
contact Shaik Mazharullah
at the National Technology Transfer Center.
Call 800/678-6882,
E-mail: smazharullah@nttc.edu
Please mention you read about it in Innovation.
