Data processing: measuring – calibrating – or testing – Measurement system in a specific environment – Earth science
Reexamination Certificate
2002-03-25
2004-07-13
Hoff, Marc S. (Department: 2857)
Data processing: measuring, calibrating, or testing
Measurement system in a specific environment
Earth science
C702S005000
Reexamination Certificate
active
06763303
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a system for classifying seafloor roughness using artificial neural network (ANN) hybrid layout from unprocessed multi-beam backscatter data. More particularly, the present invention relates to a system for online seafloor roughness classification from unprocessed multi-beam angular backscatter data using unsupervised learning as a pre-processor and supervised learning as the concluding block for improved classification, resulting in a highly efficient hybrid neural network layout to classify an unclassified dataset.
BACKGROUND OF THE INVENTION
Hitherto known neural classifier for seafloor classification [Z. Michalopoulou, D. Alexandrou, and C. de Moustier, “Application of Neural and Statistical Classifiers to the Problem of Seafloor Characterization”,
IEEE Journal of Oceanic Engineering
, Vol. 20, pp. 190-197 (1994)] describes a self-organizing map (SOM) network that is applied to multi-beam backscatter dataset. The drawback of this system is that it can use only processed data. Another drawback is its unsuitability for on-line application.
An alternate system [B. Chakraborty, R. Kaustubha, A. Hegde, A. Pereira, “Acoustic Seafloor Sediment Classification Using Self Organizing Feature Maps”,
IEEE Transactions Geoscience and Remote Sensing
, Vol. 39, No. 12, pp. 2722-2725 (2001)] describes a SOM network wherein single-beam dataset is used for seafloor classification, and this system is more suited to online use. However, a limitation of this system is that it requires pre-processing of the time-series dataset prior to classification.
In U.S. patent application Ser. No. 09/814,104 the Applicants have described a system which is incorporated in seafloor classification. This system described in this application estimates the seafloor acoustic backscattering strength with recorded root-mean-square (r.m.s) echo-voltage and the signal duration for each beam. In this system, multi-beam angular backscatter data have been acquired from the various seafloor areas around the Indian Ocean using a multi-beam acoustic system (Hydrosweep) installed onboard the Ocean Research Vessel Sagar Kanya. A drawback of the aforesaid system is that it requires large time-overhead to correct the raw data for range-related gain, seafloor slope correction, and insonification-depth normalization.
Yet another system [B. Chakraborty, H. W. Schenke, V. Kodagali, and R. Hagen, “Seabottom Characterization Using Multi-beam Echo-sounder: An Application of the Composite Roughness Theory”,
IEEE Transactions Geoscience and Remote Sensing
, Vol. 38, pp. 2419-2422 (2000)] describes a system for seafloor classification, wherein it has been observed that the seafloor roughness parameters (power-law parameters) are the ideal parameters for classification. The drawback of this system is that seafloor classification can be implemented only after carrying out physical modeling of composite roughness parameters.
OBJECTS OF THE INVENTION
The main object of the present invention is to provide a novel system for seafloor classification using artificial neural network (ANN) hybrid layout with the use of unprocessed multi-beam backscatter data.
Another object of the present invention is to provide a system for on-line (i.e., real-time) seafloor classification using backscatter data after training the self-organized mapping (SOM) network and learning vector quantization (LVQ) network.
Yet another object of the present invention is to provides a system that incorporates a hybrid network using unsupervised SOM as the first block for coarse classification of the seafloor backscatter data and supervised LVQ for highly improved performance in the said classification.
Still another object of the present invention is to provide a system which incorporates a combination of two variations of the LVQ layout to work together to achieve the best classification results.
SUMMARY OF THE INVENTION
The present invention relates to a system for classifying seafloor roughness using artificial neural network (ANN) hybrid layout from unprocessed multi-beam backscatter data. More particularly, the present invention relates to a system for online seafloor roughness classification from unprocessed multi-beam angular backscatter data using unsupervised learning as a pre-processor and supervised learning as the concluding block for improved classification, resulting in a highly efficient hybrid neural network layout to classify an unclassified dataset.
DETAILED DESCRIPTION OF THE INVENTION
Accordingly, the present invention provides a system for classifying seafloor roughness using artificial neural network (ANN) hybrid layout from unprocessed multi-beam backscatter data, said system comprising a means for generating unprocessed multi-beam backscatter r.m.s. data attached to the input of a self-organizing map (SOM) preprocessor (
20
), said SOM preprocessor being attached through one or more Learning Vector Quantization (LVQ) variants (
21
and
23
) to a memory/display module (
22
).
In an embodiment of the present invention, the means for generating unprocessed multi-beam backscatter r.m.s. data comprises a multi-beam acoustic device mounted beneath a ship's hull and attached to an r.m.s. estimator module through a beam former module.
In another embodiment of the present invention, the multi-beam acoustic device comprises a linear array of transducers connected to a roll-pitch-heave sensor through cable connection boxes and an array of transmit-receive systems.
In yet another embodiment of the present invention, the multi-beam acoustic device comprises of two identical arrays of acoustic transducers mounted at right angles to each other.
In still another embodiment of the present invention, each array of the acoustic transducer is a combination of several sub-arrays and each sub array consists of multitude of elements.
In a further embodiment of the present invention, each element form a set of channels.
In one more embodiment of the present invention, the arrays can be used either for transmission or for reception of signals.
In one another embodiment of the present invention, the multi-beam acoustic device is connected to the beam former module through a preamplifier and a time varying gain adjustment circuit.
In an embodiment of the present invention, beam forming is accomplished using appropriate delays.
In another embodiment of the present invention, the beam former module is connected to the r.m.s. estimator module through a digital to analog converter, a filter, and a analog to digital converter.
In still another embodiment of the present invention, a display means is optionally connected to the analog to digital converter.
In yet another embodiment of the present invention, the display means is connected to the analog to digital converter through a bottom-tracking gate.
In a further embodiment of the present invention, the output pattern of the r.m.s. estimator module is the envelope of the r.m.s. signal amplitude Vs. beam number in cross-track direction.
In one more embodiment of the present invention, self-organizing map (SOM) preprocessor classifies the seafloor data into various roughness types and clusters them.
In one another embodiment of the present invention, the roughness parameters are distinguished based on the ship's cross-track angular multi-beam signal backscatter shape parameter.
In an embodiment of the present invention, each cluster formed represents an unique pattern of the input data.
In another embodiment of the present invention, the number of clusters thus formed is equal to the number of differing patterns of received seafloor data set.
In yet another embodiment of the present invention, the clusters are formed with the inherent unsupervised learning feature of the SOM preprocessor.
In still another embodiment of the present invention, the clusters are formed without any prior knowledge of the number of the different types of input patterns.
In a further embodiment of the present invention, the Learning Vector Quantization (LV
Baracho Jennifer
Chakraborty Bishwajit
Joseph Anthony
Kodagali Vijay
Council of Scientific and Industrial Research
Gutierrez Anthony
Hoff Marc S.
Morgan & Lewis & Bockius, LLP
LandOfFree
System for classifying seafloor roughness does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System for classifying seafloor roughness, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System for classifying seafloor roughness will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3234631