Data Compression of One or Multidimensional Signals Utilizing an Energy Based Split Vector Quantizer via Multiple Transform Domain Representations

Technology #30041

Questions about this technology? Ask a Technology Manager

Download Printable PDF

Image Gallery
Shows the LP Coefficient Encoding Process wherein H, is the unquantized Synthesis filter response for the i'h signal frame Shows a Split Vector Quantization of LP Coefficient vector in domain j
Categories
Researchers
Wasfy Mikhael, Ph.D.
Venkatesh Krishnan
Managed By
Andrea Adkins
Assistant Director 407.823.0138
Patent Protection

Energy based split vector quantizer employing signal representation in multiple transform domains

US Patent 7,310,598 B1

Systems containing multiple transform split quantizer codebooks that will efficiently and more accurately represent data signals by measuring the extent of distortion and automatically picks which domain is better for the particular signal being transmitted or stored

Vector quantization is a powerful technique used for compressing data. The term “data” can represent signals which exist in various states (speech, images, geophysical signals, videos, etc.) which must be transformed into compact representations for the purposes of storage and transfer. “Compression” is essentially the reduction of the number of bits necessary to transmit or store these analog signals. Current technologies which utilize vector quantization for data compression employ transform domain (frequency domain) to convert the data signal into a vector. These single transform domains allocate limited bandwidth for the transmission of said vectors, thereby slowing down the process of transmission. This can also cause distortion in the recovered signal and lead to a lower reconstruction quality.

Technical Details

UCF engineering professors have designed a scheme where each signal vector is projected into multiple transform domains that will encode the signals more efficiently. Allocating transform domain for representing the signals improves upon the system’s bandwidth enabling a better recovery and reconstruction of the signal. Moreover, the invention utilizes energy-based, sub-bands in different domains which are encoded using a variety of codebooks that optimize the representation for that specific transform domain. In other words, each domain uses a coding scheme for data compression such that the coder selects transforms from the domain that best represents the signal vector. This customized selection increases transmission bandwidth, encoding and decoding speed, and reconstruction qualities producing a more efficient means, with faster data transmissions.

Benefits

  • Increased signal to noise ratio, thereby minimizing errors in the reconstructed signals
  • Increased bandwidth (data transmission) and processing capacities
  • Higher Reconstruction quality
  • Increased encoding and decoding speeds

Applications

  • Speech coding
  • Signal compression
  • Telecommunications
  • Voice over Internet Protocols (VoIP)
  • Speaker recognition