Method and apparatus for transforming between fixed-rate vector quantized data and variable rate vector quantized data
First Claim
Patent Images
1. A method comprising the steps of:
- transmitting fixed-rate vector quantized data from a first location to a second location via a first medium;
receiving the fixed-rate vector quantized data at the second location and transforming the fixed-rate vector quantized data to variable-rate vector quantized data; and
transmitting the variable-rate vector quantized data from the second location to a third location via a second medium.
3 Assignments
0 Petitions
Accused Products
Abstract
A method and apparatus is disclosed wherein fixed rate vector quantized data is transmitted from a first location to a second location via a first medium. The fixed rate vector quantized data is received at the second location and transformed to variable rate vector quantized data. Then, the transformed variable rate vector quantized data is transmitted to a third location via a second medium. At the third location, the variable rate vector quantized data may be received, and if desired, transformed back to fixed rate form. A distortion adaptive vector quantization method is employed to transform the fixed rate vector quantization data to variable rate vector quantization data.
87 Citations
53 Claims
-
1. A method comprising the steps of:
-
transmitting fixed-rate vector quantized data from a first location to a second location via a first medium; receiving the fixed-rate vector quantized data at the second location and transforming the fixed-rate vector quantized data to variable-rate vector quantized data; and transmitting the variable-rate vector quantized data from the second location to a third location via a second medium. - View Dependent Claims (2, 3)
-
-
4. Method of communicating image data from a transmitter location of television signals to at least one reception site, comprising the steps of:
-
a) receiving image data to be transmitted, and converting the image data to a plurality of multi-dimensional input vectors; b) providing at the transmitter location, a vector quantization encoder of the type that employs a first tree structured codebook to vector quantize the input vectors and sequentially transmits vector quantized data for each of the input vectors, the first codebook having a plurality N of levels of codevectors, each codevector being representative of a possible input vector, the codevectors at each successive level representing possible input vectors with greater accuracy than codevectors at a preceding level, there being an address associated with each codevector having a length corresponding to the level of the codevector in the codebook, the addresses of codevectors at a same level having a same length and the lengths increasing with each successive level; c) processing a one of the input vectors and selecting from the Nth level a codevector that most closely resembles the processed input vector, the address of the selected codevector having a length of M bits; d) transmitting to an intermediate location via a first medium, an indication of at least the M bit address associated with the selected codevector; e) repeating steps (c) and (d) for each other input vector, a plurality of the indications transmitted in step (d) representing fixed-rate vector quantized data; f) performing the following steps at the intermediate location; f1) receiving each indication transmitted in step (d) and obtaining the M bit addresses therefrom; f2) transforming each M bit address to a derivative address having a length ranging from 1 to M bits and transmitting an indication of the derivative address to at least one reception site, a plurality of the transmitted derivative address indications representing variable-rate vector quantized data. - View Dependent Claims (5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23)
-
-
24. Method of communicating image data from a transmitter location of television signals to a plurality of reception sites, comprising the steps of:
-
a) receiving image data to be transmitted, and converting the image data to a plurality of multi-dimensional input vectors; b) providing at the transmitter location, a vector quantization encoder of the type that employs a first tree structured codebook to vector quantize the input vectors and sequentially transmits vector quantized data for each of the input vectors, the first codebook having a plurality N of levels of codevectors, each codevector being representative of a possible input vector, the codevectors at each successive level representing possible input vectors with greater accuracy than codevectors at a preceding level, there being an address associated with each codevector having a length equal to L * k bits where L represents the level of the codevector in the codebook and has a value ranging from 1 to N, and where k is a function of the tree structure of the codebook, at least a portion of the address of a codevector at a given level L'"'"' of the codebook being capable of addressing a codevector at a given level P preceding level L'"'"'; c) processing a one of the input vectors and selecting from the Nth level a codevector that most closely resembles the processed input vector, the address of the selected codevector having a length equal to N * k bits; d) transmitting to an intermediate location via a first medium, an indication of at least the N * k bit address associated with the selected codevector; e) repeating steps (c) and (d) for each other input vector, a plurality of the indications transmitted in step (d) representing fixed-rate vector quantized data; f) performing the following steps at the intermediate location, there being a second tree structured vector quantization codebook at the intermediate location, the second codebook being substantially identical to the first codebook; f1) receiving a one of the indications from the transmitter location and obtaining the N * k bit address therefrom; f2) retrieving from the Nth level of the second codebook the codevector residing at the obtained N * k bit address to obtain a substantially identical representation of an associated codevector selected from the first codebook in step (c); f3) selecting an initial level of the second codebook preceding the Nth level and defining a currently selected level as the initial level and defining a value CL that identifies the currently selected level; f4) selecting the CL * k most significant bits of the obtained N * k bit address, and defining a current codevector address as the selected CL * k bits; f5) retrieving from the currently selected level of the second codebook the codevector residing at the current codevector address; f6) obtaining a measure of difference between the codevector retrieved in step (f1) and the codevector retrieved in step (f5); f7) proceeding to step (f10) only if the measure of difference is not greater than a threshold value, but otherwise selecting a next level of the second codebook succeeding the currently selected level and redefining the currently selected level as the next level, the value CL being modified to identify the currently selected level, and proceeding to step (f8); f8) proceeding to step (f9) if the currently selected level is a last level of the second codebook, but otherwise selecting the CL * k most significant bits of the N * k bit address, and redefining the current codevector address as the selected CL * k bits, and repeating steps (f5) through (f7); f9) redefining the current codevector address as the obtained N * k bit address and proceeding to step (f10); f10) defining a derivative address as the current codevector address, the derivative address having a length ranging from k to N * k bits, and transmitting an indication of at least the derivative address to at least one reception site; f11) repeating steps (f1) through (f10) for each other indication transmitted from the transmitter location in step (d), a plurality of the indications of the derivative addresses transmitted in step (f10) representing variable-rate vector quantized data. - View Dependent Claims (25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 52, 53)
-
-
41. Method of transforming fixed-rate vector quantized data to variable-rate vector quantized data wherein the fixed-rate vector quantized data comprises a plurality of M bit codevector addresses, each M bit address being associated with a codevector selected from an Nth level of a first tree structured vector quantization codebook, comprising the steps of:
-
for each M bit address; i) retrieving from the Nth level of a second codebook that is substantially identical to the first codebook the codevector residing at the M bit address to obtain a representation of the codevector selected from the M bit address of the Nth level of the first codebook; ii) selecting an initial level of the second codebook preceding the Nth level and defining a currently selected level as the initial level; iii) selecting a portion of the M bit address, the selected portion having an address length ranging from 1 to (M-1) bits and being suitable for addressing a codevector in the currently selected level of the second codebook, and defining a current codevector address as the selected portion; iv) retrieving from the currently selected level of the second codebook the codevector residing at the current codevector address; v) obtaining a measure of difference between the codevector retrieved in step (i) and the codevector retrieved in step (iv); vi) proceeding to step (ix) only if the measure of difference is not greater than a threshold value, but otherwise selecting a next level of the second codebook succeeding the currently selected level and redefining the currently selected level as the next level and proceeding to step (vii); vii) proceeding to step (viii) if the currently selected level is a last level of the codebook, but otherwise selecting another portion of the M bit address, said selected another portion having an address length ranging from 1 to (M-1) bits and being suitable for addressing a codevector in the currently selected level, and redefining the current codevector address as the selected portion and repeating steps (iv) through (vi); viii) redefining the current codevector address as the M bit address and proceeding to step (ix); ix) defining a derivative address as the current codevector address, the derivative address having a length ranging from 1 to M bits, a plurality of the derivative addresses representing variable-rate vector quantized data. - View Dependent Claims (42, 43)
-
-
44. An apparatus for transforming fixed rate vector quantized data to variable rate vector quantized data for use by operators of a television signal distribution system employing, at a transmitter location thereof, a tree search vector quantization data compressor for compressing input vectors and having a first codebook having a plurality N of levels of codevectors each representative of a possible input vector, there being an address associated with each codevector, the transmitter location transmitting for each input vector an indication of the associated address of a codevector selected from the Nth level of the first codebook that most closely resembles the input vector, the associated addresses of the selected codevectors each having a length of M bits, a plurality of the indications of the associated addresses representing fixed-rate vector quantized data, the apparatus comprising:
-
first means for receiving each indication transmitted from the transmitter location and for obtaining the M bit address therefrom; second means for transforming the obtained M bit address to a derivative address having a length ranging from 1 to M bits; and transmitter means for transmitting at least an indication of the derivative address, a plurality of transmitted indications of derivative addresses representing variable-rate vector quantized data. - View Dependent Claims (45, 46, 47, 48, 49)
-
-
50. A decoder for use by recipients of television signals employing, at a transmitter thereof, a vector quantization data compressor for compressing input vectors representative of image data to be communicated to the recipients, and having a first codebook having plural levels of codevectors representative of possible input vectors, there being an address associated with each codevector, the transmitter transmitting variable-rate vector quantized data comprising data indicative of the addresses of codevectors selected from the first codebook that most closely resemble the input vectors, the addresses of the selected codevectors having variable lengths, the decoder comprising:
-
a second codebook being substantially identical to the first codebook; first means for receiving the variable-rate vector quantized data and for obtaining the variable length addresses therefrom; second means for retrieving from the second codebook the codevector residing at each obtained address; converter means for converting the retrieved codevectors to NTSC image data and for providing the NTSC image data at an output thereof for substantially reproducing images input to the transmitter; and
,transformer means for optionally transforming the variable-rate vector quantized data received by the first means to fixed-rate vector quantized data when it is desired to store vector quantized image data on magnetic media in the vicinity of the decoder. - View Dependent Claims (51)
-
Specification