Method of quantizing predictive errors

• US 4,815,078 A
• Filed: 03/31/1987
• Issued: 03/21/1989
• Est. Priority Date: 03/31/1986
• Status: Expired due to Term

• Current Assignee: Fuji Photo Film Co Limited (Fujifilm Holdings Corporation)
• Original Assignee: Fuji Photo Film Co Limited (Fujifilm Holdings Corporation)
• Inventors: Shimura, Kazuo
• Primary Examiner(s): ATKINSON, CHARLES
• Application Number:
  US07/033,441
• Time in Patent Office: 721 Days
• Field of Search: 371/30, 371/72, 382/56, 382/57, 358/133, 358/135, 358/260, 375/26, 375/27
• US Class Current: 714/746
• CPC Class Codes: H04N 19/50 using predictive coding H04N19/61 takes precedence
• CPC Class Codes: Y10S 250/909 Methods and apparatus ancillary to stimulable phosphor systems

Abstract

Predictive errors are grouped into a plurality of quantizing ranges, and the predictive errors in the respective quantizing ranges are represented by representative quantizing values assigned respectively to the quantizing ranges. The predictive error data items in each of the quantizing ranges are processed to provide a weighted mean value based on the frequency distribution thereto. The weighted mean value or a value close thereto is used as a representative quantized value for that quantizing range.

Claims

1. A method of compressing digital image data to be stored or transmitted, the digital image data representing the image information at corresponding pixels of an image, comprising the steps of:

determining for each pixel a predicted image data value,

determining for each predicted image data value a predictive error value;

assigning the predictive error values to groups, each group consisting of a selected plurality of predictive error values;

determining the frequency of occurrence of each predictive error value in each of said groups;

determining a weighted mean value for the predictive error values in each of said groups based on the predictive error values and their frequencies; and

selecting the weighted mean value of the group for each predictive error value within the group as the representative quantized predictive error value of the corresponding pixel.

2. The method of compressing digital image data as claimed in claim 1 wherein said step of determining a weighted mean value comprises the steps of calculating a weighted mean value based on the frequency distribution of the predictive error values within the group, determining if the calculated weighted mean value is an integer, and selecting as the weighted mean value of the group the calculated weighted mean value when it is an integer and an integer which is close to the calculated weighted mean value when it is not an integer.

3. The method of compressing digital image data as claimed in claim 1 wherein said step of determining a weighted mean value comprises the steps of calculating a weighted mean value based on the frequency distribution of the predictive error values within the group according to the equation:

##EQU4## where;

x=predictive error

f(x)=frequency of predictive error x

x_i -x_1+i =quantizing range of G group

Xi=representative quantized predictive error value (weighted mean value).

4. A method of compressing digital image data values corresponding to the image information of pixels comprising an image, comprising the steps of:

determining for each of said pixels a predictive error value corresponding to the difference between the actual image data value and a predicted image value for the pixel;

grouping the predictive error values in a plurality of groups according to quantizing ranges;

determining the frequency distribution of each of the predictive error values in the groups;

calculating a weighted mean value for each group of predictive error values according to the frequency distribution; and

outputting for each pixel the calculated weighted mean value of the group comprising the corresponding predictive error value of the pixel.