Method and apparatus for the automatic generation of subject to background transition area boundary lines and subject shadow retention
First Claim
1. A method for compositing motion picture and video images by automatically generating inner and outer boundary lines on either side of a subject-to-background transition area comprising the steps of:
- a) storing in a first memory the signal levels of each signal channel comprising the signals at each pixel location in a clear image frame with a subject absent, b) storing in a second memory the signal levels of each signal channel comprising the signals at each pixel location in a current image frame with said subject in place, c) subtracting the signal levels of pixels in said clear image frame from signal levels of corresponding pixels in said current image frame to obtain their difference, d) storing in a third memory a 0.0 at all pixel addresses where said difference approximates zero, and storing a 1.0 for all remaining pixels to form a field of zeros in the background area and a field of ones in the subject and transition areas, e) defining an outer boundary line as the addresses of those pixels forming a continuous line of pixels having been assigned a 0.0 and located adjacent to pixels having been assigned a 1.0, thereby forming an outer boundary line on the background adjacent to the edge of the subject-to-background transition area, f) shrinking said outer boundary line inward to form an inner boundary line on the subject at a sufficient distance from said subject'"'"'s edge to avoid including elements of the background, thereby automatically generating said inner and outer boundary lines.
1 Assignment
0 Petitions
Accused Products
Abstract
Changes in a current image frame from variations in room light level, and automatic camera adjustments, are determined and incorporated into a clear (reference) frame, thereby resulting in a zero difference in the background region when the clear frame and a current frame are subtracted. Any deviation from zero defines the beginning edge of a transition from background to subject. The location of the outer edge of a transition area is further refined by subtracting derivatives of the RGB signals in each frame, and by the use of a fourth color channel when available. The difference of derivitives (dcu-dcl) identifies the subject to background transition area, whose edges define the inner and outer boundary lines. Shadow areas on the backing are retained and may be transferred to the inserted background scene. Small areas on the backing having the same color and luminance as the subject are prevented from printing through the subject by a second “derivative” alpha channel for the small areas.
-
Citations
19 Claims
-
1. A method for compositing motion picture and video images by automatically generating inner and outer boundary lines on either side of a subject-to-background transition area comprising the steps of:
-
a) storing in a first memory the signal levels of each signal channel comprising the signals at each pixel location in a clear image frame with a subject absent, b) storing in a second memory the signal levels of each signal channel comprising the signals at each pixel location in a current image frame with said subject in place, c) subtracting the signal levels of pixels in said clear image frame from signal levels of corresponding pixels in said current image frame to obtain their difference, d) storing in a third memory a 0.0 at all pixel addresses where said difference approximates zero, and storing a 1.0 for all remaining pixels to form a field of zeros in the background area and a field of ones in the subject and transition areas, e) defining an outer boundary line as the addresses of those pixels forming a continuous line of pixels having been assigned a 0.0 and located adjacent to pixels having been assigned a 1.0, thereby forming an outer boundary line on the background adjacent to the edge of the subject-to-background transition area, f) shrinking said outer boundary line inward to form an inner boundary line on the subject at a sufficient distance from said subject'"'"'s edge to avoid including elements of the background, thereby automatically generating said inner and outer boundary lines. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18)
-
-
17. The method of claim 14 in which said clear (cl) frame is multiplied by the ratio (I cu/Icl) to form a corrected clear frame that includes shadows extending behind a subject'"'"'s transition area, and said corrected clear frame constitutes a background reference frame.
-
18. The method of claim 1 in which the subtraction in step 1.c) and the assignments in step 1.d) are replaced by alternate methods of comparing the clear frame and the current frame to determine the foreground region, such as by computing a preliminary alpha and processed foreground and assigning a 1.0 to any pixel for which the preliminary processed foreground exceeds some low threshold, or for which the preliminary alpha is below another low threshold, and assigning a 0.0 to all other pixels, and where said preliminary alpha and said computation of said processed foreground and preliminary alpha comprise the following steps,
a) estimating the similarity of colors between corresponding pixels in the clear frame and the current frame, on a scale ranging from 0.0 to 1.0, b) subtracting the product of the preliminary alpha and the clear frame color from the current frame color, forming said preliminary processed foreground.
-
19. A system for compositing motion picture and video images by automatically generating inner and outer boundary lines on either side of a subject-to-background transition area comprising:
-
a) a first memory for storing the signal levels of each signal channel comprising the signals at each pixel location in a clear image frame with a subject absent, b) a second memory for storing the signal levels of each signal channel comprising the signals at each pixel location in a current image frame with said subject in place, c) a comparator coupled to said first memory and said second memory for subtracting the signal levels of pixels in said clear image frame from signal levels of corresponding pixels in said current image frame to obtain their difference, d) a third memory coupled to said comparator for storing a 0.0 at all pixel addresses where said difference approximates zero, and storing a 1.0 for all remaining pixels to form a field of zeros in the background area and a field of ones in the subject and transition areas, e) means for defining an outer boundary line as the addresses of those pixels forming a continuous line of pixels having been assigned a 0.0 and located adjacent to pixels having been assigned a 1.0, thereby forming an outer boundary line on the background adjacent to the edge of the subject-to-background transition area, f) means for shrinking said outer boundary line inward to form an inner boundary line on the subject at a sufficient distance from said subject'"'"'s edge to avoid including elements of the background, thereby automatically generating said inner and outer boundary lines.
-
Specification