Differential PCM video transmission system employing horizontally offset five pixel groups and delta signals having plural non-linear encoding functions

• US 4,661,862 A
• Filed: 04/27/1984
• Issued: 04/28/1987
• Est. Priority Date: 04/27/1984
• Status: Expired due to Fees

• Current Assignee: RCA Corporation
• Original Assignee: RCA Corporation
• Inventors: Thompson, Charles R.
• Primary Examiner(s): Cardillo, Raymond F.
• Assistant Examiner(s): Young, Wayne R.
• Application Number:
  US06/604,501
• Time in Patent Office: 1,096 Days
• Field of Search: 358/335, 358/133, 358/135, 358/141, 358/263, 358/13, 358/12, 358/310, 358/11, 358/137, 358/140, 358/138, 375/26-27, 375/122, 360/9.1, 360/32, 360/33.1
• US Class Current: 386/328
• CPC Class Codes: H04N 11/046 DPCM
• CPC Class Codes: H04N 11/28 involving bandwidth reduction, e.g. subsampling
• CPC Class Codes: H04N 19/593 involving spatial prediction techniques
• CPC Class Codes: H04N 19/90 using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals

Abstract

A wide-aspect-ratio HDTV signal having 1600 luminance pixels/32 μS line and 980 lines/frame is data reduced by deleting odd pixels of each line of TV field #1 and even pixels of each line of TV field #2 to produce 800 pixels per line, and is further data compressed by a form of differential-pulse-code modulation. The DPCM organizes the pixels into groups of five, and for each group transmits or records one reference pixel at full N-bits. The pixels to the right and left of the reference pixel are compared therewith to produce difference signals of M bits, where M<N. The remaining two pixels of each group are compared with reference pixels from the lines above and below to form M-bit difference signals. The difference signals are transmitted. The DPCM coding reduces the 800 pixels/line to the equivalent of 480 pixels/line. The color portion of the wide-aspect-ratio HDTV is similarly coded to produce the equivalent of 240 pixel/line. The 480 luminance samples and 240 color samples are added to form 720 samples/line, or 1440 samples/64 μsec which can be accommodated by a CCIR-standard-601 signal processing system.

Claims

1. A method for compressing digital data in the form of time-sequential television pixels representing successive raster lines, each pixel including N bits comprising the recurrent steps of:

organizing said pixels into recurrent groups of first, second, third, fourth and fifth time-sequential pixels, each group of pixels of a given raster line being offset horizontally by two pixel positions with respect to a group of pixels of a preceding line;

transmitting each second pixel of each group with a resolution of N-bits;

comparing said first and third pixels with said second pixel located therebetween for each said group to produce first and second N/2 bit delta signals representing the difference between said second pixel and said first and third pixels, respectively;

transmitting each said N/2-bit first delta signal of a given group before transmitting said second pixel of said given group;

transmitting each said N/2-bit second delta signal of said given group after transmitting said second pixel of said given group;

comparing said fourth pixel of said given group with said second pixel having a corresponding horizontal position on one of the next upper and next lower horizontal lines to produce an N/2-bit third delta signal representing the difference between said fourth and second vertically adjacent pixels;

transmitting said third delta signal immediately after said second delta signal of said given group;

comparing said fifth pixel of said given group with said second pixel having a corresponding horizontal position on the other of said next upper and next lower horizontal lines to produce an N/2-bit fourth delta signal representing the difference between said fifth and second vertically adjacent pixels; and

transmitting said fourth delta signal immediately after said third delta signal of said given group.

2. A method according to claim 1 wherein N equals 8 and N/2 equals 4.

3. A method according to claim 1 wherein the values of N/2-bit delta signals are nonlinearly related to the magnitude of the differences which they represent.

4. A method for compressing digital data in the form of time-sequential television pixels representing successive raster lines, each pixel including N bits comprising the recurrent steps of:

organizing said pixels into recurrent groups of first, second, third, fourth and fifth time-sequential pixels, each group of pixels of a given raster line being offset horizontally by two pixel positions with respect to a group of pixels of a preceding line

transmitting each second pixel of each group with a resolution of N-bits;

comparing said first and third pixels with said second pixel located therebetween for each said group to produce first and second M-bit delta signals representing the difference between said second pixel and said first and third pixels, respectively;

where M is a number of bits less than N;

transmitting each said M-bit first delta signal of a given group before transmitting said second pixel of said given group;

transmitting each said M-bit second delta signal of said given group after transmitting said second pixel of said given group;

comparing said fourth pixel of said given group with said second pixel having a corresponding horizontal position on one of the next upper and next lower horizontal lines to produce an M-bit third delta signal representing the difference between said fourth and second vertically adjacent pixels;

transmitting said third delta signal immediately after said second delta signal of said given group;

comparing said fifth pixel of said given group with said second pixel having a corresponding horizontal position on the other of said next upper and next lower horizontal lines to produce an M-bit fourth delta signal representing the difference between said fifth and second vertically adjacent pixels; and

transmitting said fourth delta signal immediately after said third delta signal of said given group.

5. A method for compressing digital picture information, comprising:

transmitting a first pixel-representative word at full bit depth;

comparing a second pixel-representative word with said first pixel-representative word to form a difference word representative of the amplitude difference between said first and second words;

encoding said difference word to form a delta word having less than said full bit depth, said encoding comprising subjecting said difference word to a plurality of non-linear encoding transfer functions, each said transfer function having a first region representing small amplitude increments when said amplitude difference represented by said difference word is small and having at least one second region representing large amplitude differences when said amplitude difference represented by said difference signal is large and

selecting a given one of said transfer functions for encoding said difference word as a function of the amplitude of said first pixel-representative word to thereby provide an encoded delta word having greater accuracy for small differences over a range of values of said amplitude of said first pixel representative word.

6. A method for transmitting N-bit digital television representative signals through a transmission path, comprising:

organizing said digital signals into recurrent groups of five time-adjacent samples representative of horizontally-adjacent pixels on each one of a plurality of lines of said television representative signal;

offsetting each group of samples by two pixel positions from line-to-line;

comparing first and second samples of a group of a first one of said lines with a second sample of said group to form first and second difference signals, respectively;

comparing a fourth sample of said group with the second sample of a group on a second horizontal line adjacent said first horizontal line to form a third difference signal;

comparing a fifth sample of said group with the second sample of a group on a third horizontal line adjacent said first horizontal line to produce a fourth difference signal;

representing each of said first, second, third and fourth difference signals by words of M bits, where M is less than N bits;

transmitting said first, second, third and fourth difference signals of M bits and said N-bit second signal.

7. A method according to claim 6 wherein said N-bit words have 8 bits.

8. A method according to claim 6 wherein said M-bit words have 4 bits.

9. A method according to claim 6 wherein said transmission path comprises video record means.

10. A method for reconstituting digital picture information encoded in a format in which a first pixel is transmitted in unencoded form as an N-bit word and a second pixel is transmitted in an encoded form as an M-bit delta word, M being less than N, said delta word being encoded as given one of a plurality of non-linear functions of an amplitude difference between said first and second pixels, said given non-linear function being selected in accordance with the amplitude of said N-bit word, comprising:

receiving said N-bit pixel and M-bit delta words;

selecting one of a plurality of decoding transfer functions in accordance with a given number of most significant bits of said N-bit pixel word;

applying said M-bit delta word to said transfer function selected by said N-bit word to regenerate said amplitude difference between said first and second pixels; and

combining said N-bit word with said regenerated amplitude difference to reconstitute said second pixel with a resolution of N-bits.