Luminance transition coding method for software motion video compression/decompression

• US 5,465,118 A
• Filed: 12/17/1993
• Issued: 11/07/1995
• Est. Priority Date: 12/17/1993
• Status: Expired due to Fees

• Current Assignee: Lenovo Singapore Pte Limited (Lenovo Group Ltd.)
• Original Assignee: International Business Machines Corporation
• Inventors: Hancock, Steven M., Pietras, Mark A.
• Primary Examiner(s): Chin, Tommy P.
• Assistant Examiner(s): RAO, ANAND SHASHIKANT
• Application Number:
  US08/170,044
• Time in Patent Office: 690 Days
• Field of Search: 348/395, 348/396, 348/394, 348/397, 348/392, 348/391, 348/631
• US Class Current: 348/396.1
• CPC Class Codes: H04N 19/94 Vector quantisation
• CPC Class Codes: H04N 19/98 Adaptive-dynamic-range coding [ADRC]

Abstract

An digital video image processing system employs codes of various lengths to encode the luminance information within the digital video images. The inventive method and apparatus divides a video image into non-overlapping regions containing a plurality of pixels with luminance values. Within each region, the method computes the total luminance variation across the region, the direction of greatest luminance change within the region, the base, or lowest, luminance value within the region and the location and type of luminance transition within the region. Depending on the variation of the luminance values within a region and on the quality desired in the decoded image, the computed information is used to encode the luminance values in short, medium, or long codes. The luminance codes may be predetermined and unchanging or they may vary from image to image or within given images.

Claims

1. An image processing system which compresses a digitized video image comprising:

A. means for partitioning the image into non-overlapping contiguous regions containing a plurality of pixels each having a unique luminance value indicative of the luminance of its associated pixel; and

B. means for compressing the luminance information of pixels in each region by encoding the luminance values within each region into a compressed code, wherein said means for compressing includes;

1. means for comparing the luminance values of the pixels located along the perimeter of a region to determine a pixel containing the smallest luminance value and its location within the region and for providing a base luminance value indicative of the smallest luminance value;

2. means for determining direction of greatest luminance change from the base luminance value location;

3. means for computing a total change in luminance along the direction of greatest luminance change starting from the base luminance value location;

4. means for selecting a transition function value indicative of the type of luminance transition within the region starting from the location of the base luminance value wherein the luminance transition values are indicative of the following functions;

a. a linear luminance transition;

b. a low contrast luminance edge transition at one quarter of the distance across the region;

c. a low contrast luminance edge transition at one half of the distance across the region;

d. a low contrast luminance edge transition at three quarters of the distance across the region;

e. a high contrast luminance edge transition at one quarter of the distance across the region;

f. a high contrast luminance edge transition at one half of the distance across the region; and

g. a high contrast luminance edge transition at three quarters of the distance across the region; and

3. The system of claim 1 wherein:

A. the linear luminance transition exhibits equal luminance steps across the region;

B. the low contrast luminance edge transition exhibits unequal luminance steps across the region, the luminance edge transition being characterized by a variation in luminance along a line perpendicular to the direction to greatest luminance change across the region, the diffuse nature of which is represented by an error diffusion distribution in the direction to greatest luminance change across the region; and

C. the high contrast luminance edge transition exhibits unequal luminance steps across the region, the luminance edge transition being characterized by a relatively sharp variation in luminance along a line perpendicular to the direction of greatest luminance change across the region.

4. The image processing apparatus of claims 1, wherein the codes are predetermined but may change from image to image or within a given image.

5. The image processing apparatus of claim 1, wherein the system includes conversion equipment to convert video signals from a three-color or composite-video format to a luminance, chrominance video format.

7. The method of compressing of claim 5 wherein the luminance transition array contains data which does not vary from image to image or from region to region within an image.

8. The method of compressing of claim 5 further comprising employing a predetermined set of transition functions which are modified between images or from region to region within an image.

2. means for generating a digitized encoded data stream containing encoded data indicative of the direction of greatest luminance change and an index value representative of the luminance characteristics within the region, wherein the index value is selected as a function of the base luminance value, the direction of greatest luminance change, the total change in luminance value, and the transition function value wherein said means for generating a digitized encoded data stream, further comprises:

means for generating a short code of compressed video image data comprising i) M₁ number of bits indicative of the direction of greatest luminance change, and ii) M₂ number of bits indicative of the base luminance value;

means for generating a medium length code of compressed video image data comprising i) M₃ number of bits indicative of the direction of greatest luminance change, ii) M₄ number of bits indicative of an index value which is a pointer to an array containing encoded luminance information; and

means for providing said encoded data stream as said short length code when the value of total variation in luminance is less than a first threshold value, and for providing said encoded data stream as said medium length code when the value of the total variation in luminance is not less then the first threshold value.

6. A method of compressing the amount of data required to represent the luminance information of a digitized image which is represented in a luminance chrominance format, comprising the steps:

A. partitioning the image into non-overlapping contiguous regions each containing a plurality of luminance values and selecting a region for processing;

B. determining a base luminance value from luminance values located along the perimeter of the region and determining a base luminance location indicative of the location of the base luminance value in the region;

C. determining a direction in the region having a greatest luminance change from the base luminance location;

D. determining a total change in luminance within the region by determining the largest luminance difference between the base luminance value and the remaining luminance values in the region;

E. determining a type of luminance transition within the region from the location of the base luminance value, and selecting a luminance transition function value indicative of the type of luminance transition from a set of possible luminance transition functions; and

F. encoding the luminance information for the region based on the base luminance value, the total change in luminance, the direction of greatest luminance change, and the luminance transition function value indicative of the type of luminance transition within the region, and providing an encoded image signal, wherein the step of encoding further comprises the steps of;

a. encoding the region'"'"'s luminance data in a short code consisting of two bits which indicate the direction of greatest luminance change within the region and six bits which indicate the region'"'"'s base luminance value when the total change in luminance throughout the region is less than a first threshold value; and

b. encoding the region'"'"'s luminance data in a medium-length code consisting of four bits which indicate the direction of greatest luminance change within the region and twelve bits indicative of the an index value which is a pointer to a luminance transition array containing encoded region luminance information, when the total change in luminance across the region is not less than the first threshold value.

9. An image processing system which compresses a digitized video image comprising:

A. means for partitioning the image into non-overlapping contiguous regions containing a plurality of pixels each having a unique luminance value indicative of the luminance of its associated pixel; and

B. means for compressing the luminance information of pixels in each region by encoding the luminance values within each region into a compressed code, wherein said means for compressing includes

1. means for comparing the luminance values of the pixels located along the perimeter of a region to determine a pixel containing the smallest luminance value and its location within the region, and for providing a base luminance value indicative of the smallest luminance value;

2. means for determining a direction of greatest luminance change from the base luminance value location;

3. means for computing a total change in luminance along the direction of greatest luminance change starting from the base luminance value location;

4. means for selecting a transition function value indicative of the type of luminance transition within the region starting from the location of the base luminance value; and

5. means for generating a digitized encoded data stream containing encoded data indicative of the direction of greatest luminance change and an index value representative of the luminance characteristics within the region, wherein the index value is selected as a function of the base luminance value, the direction of greatest luminance change, the total change in luminance value, and the transition function value, wherein said means for generating a digitized encoded data stream, further comprises;

means for generating a short code of compressed video image data comprising i) two bits indicative of the direction of greatest luminance change, and ii) six bits indicative of the base luminance value;

means for generating a medium length code of compressed video image data comprising i) four bits indicative of the direction of greatest luminance change, ii) twelve bits indicative of an index value which is a pointer to an array containing encoded luminance information; and

means for providing said encoded data stream as said short length code when the value of total variation in luminance is less than a first threshold value, and for providing said encoded data stream as said medium length code when the value of the total variation in luminance is not less then the first threshold value.

10. An image processing system which compresses a digitized video image, the image processing system comprising:

A means for partitioning the image into non-overlapping contiguous regions containing a plurality of pixels each having a unique luminance value indicative of the luminance of its associated pixel; and

B. means for compressing the luminance information of pixels in each region by encoding the luminance values within each region into a compressed code, wherein said means for compressing includes

1. means for comparing the luminance values of the pixels located along the perimeter of a region to determine a pixel containing the smallest luminance value and its location within the region, and for providing a base luminance value indicative of the smallest luminance value;

2. means for determining a direction of greatest luminance change from the base luminance value location;

13. The method of claim 10 wherein following the step of comparing each of the error values to determine the smallest error value, the method of compressing performs a step of comparing the smallest error value against a predetermined quality threshold value to determine if sufficient image quality has been retained in the encoded image signal.

11. means for computing a total change in luminance along the direction of greatest luminance change starting from the base luminance value location;

4. means for selecting a transition function value indicative of the type of luminance transition within the region starting from the location of the base luminance value; and

5. means for generating digitized encoded data stream containing encoded data indicative of the direction of greatest luminance change and an index value representative of the luminance characteristics within the region, wherein the index value is selected as a function of the base luminance value, the direction of greatest luminance change, the total change in luminance value, and the transition function value wherein said means for generating a digitized encoded data stream, further comprises;

means for generating a short code of compressed video image data comprising i) M₁ number of bits indicative of the direction of greatest luminance change, and ii) M₂ number of bits indicative of the base luminance value;

means for generating a medium length code of compressed video image data comprising i) M₃ number of bits indicative of the direction of greatest luminance change, ii) M₄ number of bits indicative of an index value which is a pointer to an array containing encoded luminance information; and

means for providing said encoded data stream as said short length code when the value of total variation in luminance is less than a first threshold value, and for providing said encoded data stream as said medium length code when the value of the total variation in luminance is not less then the first threshold value wherein said means for generating a digitized encoded data stream, further comprises means for generating a long length code of four bytes of compressed image data.

14. The method of claim 11 wherein if the smallest error value is less than the predetermined quality threshold value, the encoded image signal is encoded as a two byte signal.

12. A method of compressing the amount of data required to represent the luminance information of a digitized image which is represented in a luminance chrominance format, comprising the steps:

A. partitioning the image into non-overlapping contiguous regions each containing a plurality of luminance values and selecting a region for processing;

B. determining a base luminance value from luminance values located along the perimeter of the region and determining a base luminance location indicative of the location of the base luminance value in the region said selecting a base luminance value step comprising the steps of;

a. comparing the luminance values of all the pixels located on the perimeter of the region;

b. determining the pixel with a smallest luminance value based on the comparison step; and

c. assigning the smallest luminance value to the base luminance value;

C. determining a direction in the region having a greatest luminance change from the base luminance location;

D. determining a total change in luminance within the region by determining the largest luminance difference between the base luminance value and the remaining luminance values in the region;

E. determining a type of luminance transition within the region from the location of the base luminance value, and selecting a luminance transition function value indicative of the type of luminance transition from a set of possible luminance transition functions; and

F. encoding the luminance information for the region based on the base luminance value, the total change in luminance, the direction of greatest luminance change, and the luminance transition function value indicative of the type of luminance transition within the region, and providing an encoded image signal wherein the step of encoding further comprises the steps of;

a. computing a plurality of average luminance values each representative of the average luminance for an associated one of a plurality of pixel lines which each run within the region in a direction perpendicular to the direction of greatest luminance change;

b. for each of the possible luminance transition functions each having a unique luminance transition function value

1) adding the base luminance value to the total change in luminance value, and adding to a resultant sum the luminance transition function value to form a look-up table value;

2) retrieving from an array, as a function of the look-up table value, a plurality of test compression values;

3) generating an error value indicative of the luminance difference between a test compressed region, as characterized by the test compression values, and the original uncompressed region;

c. comparing each of the error values with one another to determine a smallest error value; and

d. identifying a selected luminance transition function value associated with the smallest error value.

15. The method of claim 12 wherein if the lowest error value is not less than the predetermined quality threshold value, the encoded signal is encoded as a four byte signal.

16. A method of compressing the amount of data required to represent the luminance information of a digitized image which is represented in a luminance chrominance format, comprising the steps of:

A. partitioning the image into non-overlapping contiguous regions each containing a plurality of luminance values and selecting a region for processing;

B. determining a base luminance value from luminance values located along the perimeter of the region and determining a base luminance location indicative of the location of the base luminance value in the region;

C. determining a direction in the region having a greatest luminance change from the base luminance location;

D. determining a total change in luminance within the region by determining the largest luminance difference between the base luminance value and the remaining luminance values in the region;

E. determining a type of luminance transition within the region from the location of the base luminance value, and selecting a luminance transition function value indicative of the type of luminance transition from a set of possible luminance transition functions; and

F. encoding the luminance information for the region based on the base luminance value, the total change in luminance, the direction of greatest luminance change, and the luminance transition function value indicative of the type of luminance transition within the region, and providing an encoded image signal wherein said encoding step further comprises the steps of;

1. encoding the region'"'"'s luminance data in a short code consisting of m1 number bits which indicate the direction of greatest luminance change within the region and m2 number of bits which indicate the region'"'"'s base luminance value when the total change in luminance throughout the region is less than a first threshold value; and

2. encoding the region'"'"'s luminance data in a medium-length code consisting of m3 number of bits which indicate the direction of greatest luminance change within the region and m4 number of bits indicative of an index value which is a pointer to a luminance transition array containing encoded region luminance information, when the total change in luminance across the region is not less than the first threshold value and further comprising the step of encoding the region'"'"'s luminance data in a long length code comprising several bits indicative of the direction of greatest luminance change within the region, and a plurality of bits indicative of at least four luminance values which together characterize the luminance transition within the region when the total change in luminance across the region is not less than both the first threshold value and a second threshold value.

17. A method of compressing the amount of data required to represent the luminance information of a digitized image which is represented in a luminance chrominance format, comprising the steps:

A. partitioning the image into non-overlapping contiguous regions each containing a plurality of luminance values and selecting a region for processing;

B. determining a base luminance value from luminance values located along the perimeter of the region and determining a base luminance location indicative of the location of the base luminance value in the region;

C. determining a direction in the region having a greatest luminance change from the base luminance location;

D. determining a total change in luminance within the region by determining the largest luminance difference between the base luminance value and the remaining luminance values in the region;

E. determining a type of luminance transition within the region from the location of the base luminance value, and selecting a luminance transition function value indicative of the type of luminance transition from a set of possible luminance transition functions; and

F. encoding the luminance information for the region based on the base luminance value, the total change in luminance, the direction of greatest luminance change, and the luminance transition function value indicative of the type of luminance transition within the region, and providing an encoded image signal wherein the set of possible luminance transition functions comprises transition functions representative of;

1. a linear luminance transition;

2. a low contrast luminance edge transition at one quarter of the distance across the region;

3. a low contrast luminance edge transition at one half of the distance across the region;

4. a low contrast luminance edge transition at three quarters of the distance across the region;

5. a high contrast luminance edge transition at one quarter of the distance across the region;

6. a high contrast luminance edge transition at one half of the distance across the region; and

7. a high contrast luminance edge transition at three quarters of the distance across the region.