Technique for reducing transmission error propagation in an adaptively predicted DPCM video encoder

• US 4,255,763 A
• Filed: 12/26/1979
• Issued: 03/10/1981
• Est. Priority Date: 12/26/1979
• Status: Expired due to Term

• Current Assignee: Bell Telephone Laboratories, Inc. (Nokia Corporation)
• Original Assignee: Bell Telephone Laboratories, Inc. (Nokia Corporation)
• Inventors: Maxemchuk, Nicholas F., Stuller, John A.
• Primary Examiner(s): Britton, Howard W.
• Application Number:
  US06/107,262
• Time in Patent Office: 440 Days
• Field of Search: 358/135, 358/133, 358/138, 358/260, 364/515
• US Class Current: 375/240.12
• CPC Class Codes: H04N 19/593 involving spatial prediction techniques

Abstract

Propagation of transmission errors in adaptively predicted, DPCM encoded video signals is reduced by including a leak circuit (114) in the encoder (FIG. 1). The leak circuit modifies each prediction in response to local characteristics in the area of the picture being encoded.

Claims

1. Apparatus for encoding a video signal representing the intensity of picture elements (pels) of a picture including:

means (105) for predicting the intensity of each pel as a function of reconstructed intensity values of previously encoded pels;

means (107) for forming an error signal indicative of the difference between the intensity of each pel and the predicted intensity thereof;

means (108) for quantizing said error signal; and

means (110) for combining the quantized error signal with said predicted intensity to form said reconstructed intensity value,

wherein said predicting means (105) includes;

first means (113) for forming an intermediate prediction of the intensity of said each pel as a function of the reconstructed intensity value of at least a selected one of the pels in the neighborhood thereof; and

second means (114) for forming said predicted intensity as a function of said intermediate prediction,

characterized in that said second forming means (114) forms said predicted intensity also as a function of the reconstructed intensity values of at least first and second ones of said neighboring pels.

2. The invention defined in claim 1 wherein said predicting means (105) further includes means (111) for storing reconstructed intensity values for pels in said neighborhood of said each pel, and

said first and second means (113 and 114) are each responsive to said storing means (111).

3. The invention defined in claim 2 wherein said intermediate prediction (x_i,j) is modified to form said predicted intensity (p_i,j) in accordance with:

space="preserve" listing-type="equation">p.sub.i,j =α

β

x.sub.i,j +α

(1-α

)F(y)+(1-α

)η

where α and

β

are scaling constants 0<

α

<

1, 0<

β

<

1, η

is a preselected constant derived from the maximum range of intensity of said elements, and F(y) is the local intensity in the neighborhood of said each pel.

4. The invention defined in claim 2 wherein said second forming means includes:

(a) an average circuit (245) for deriving the local intensity in the neighborhood of said each pel,

(b) a first attenuator (255) for forming the sum of (1) said intermediate prediction attenuated by a factor β

, and (2) said local intensity attenuated by a factor 1-β

, where 1 <

β

<

1, and

(c) a second attenuator (265) for forming the sum of (1) the output of said first attenuator attenuated by a factor α

, and (2) a predetermined constant η

attenuated by a factor 1-α

, where 0<

α

<

1 and η

is near the middle of the range of intensity values of said pels.

5. Apparatus for encoding a video signal comprising a series of samples each representing the intensity of an associated element of a picture, comprising:

(a) means for predicting the value of each sample based upon the values of prior samples;

(b) means for modifying each prediction to reduce the effects of transmission errors on the decoded version of said video signal, and

(c) means for transmitting the error signal formed by subtracting each modified prediction from the corresponding one of said samples,

wherein said modifying means includes means for attenuating said prediction as a function of (1) the intensity of ones of said picture elements in the vicinity of the element associated with the prediction and (2) the range of possible intensity values for said picture elements.

6. The invention defined in claim 5 wherein said attenuating means is arranged to provide a modified prediction

7. The invention defined in claim 6 wherein said apparatus also includes:

(1) means for forming reconstructed versions of each sample using outputs from said transmitting means and said modifying means, and

(2) means for storing ones of said reconstructed versions,

wherein said predicting means and said modifying means are each responsive to said storing means.

8. A method of encoding a video signal representing the intensity of picture elements (pels) of a picture including the steps of:

predicting the intensity of each pel as a function of reconstructed intensity values of previously encoded pels;

forming an error signal indicative of the difference between the intensity of each pel and the predicted intensity thereof;

quantizing said error signal; and

combining the quantized error signal with said predicted intensity to form said reconstructed intensity value,

wherein said predicting step includes the steps of;

forming an intermediate prediction of the intensity of said each pel as a function of the reconstructed intensity value of at least a selected one of the pels in the neighborhood thereof; and

forming said predicted intensity as a function of said intermediate prediction,

characterized in that said last mentioned forming step includes forming said predicted intensity also as a function of the reconstructed intensity values of at least first and second ones of said neighboring pels.