Self-adapting method and device for the inlaying of color video images

• US 5,194,941 A
• Filed: 06/06/1991
• Issued: 03/16/1993
• Est. Priority Date: 10/06/1989
• Status: Expired due to Term

• Current Assignee: Thomson Video Equipment
• Original Assignee: Thomson Video Equipment
• Inventors: Galichet, Sylvie, Grimaldi, Jean-Luc
• Primary Examiner(s): Groody, James J.
• Assistant Examiner(s): Metjahic, Safet
• Application Number:
  US07/700,200
• Time in Patent Office: 649 Days
• Field of Search: 358/22 CK, 358/22 PIP, 358/22 C, 358/22, 358/182, 358/183, 358/181, 382/16, 382/22, 395/128, 395/129
• US Class Current: 348/587
• CPC Class Codes: H04N 5/2622 Signal amplitude transition in the zone between image portions, e.g. soft edges
• CPC Class Codes: H04N 9/75 Chroma key

Abstract

The method for the inlaying of color video images consists on the one hand in generating a switching signal, or dividing key, applying a gain on the image points belonging to the contour of the subject to be inlaid, in proportion to the distance value (DIST) between the color of the point and that of the back-ground, and on the other hand in carrying out a mixing of the source image and a new background image which is a function of the value of the dividing key signal at the point in question and such that the values of the source signal and of the new background signal are fixed at constants over the width of the contour.

Claims

1. A method for inlaying color video images, including the generation of a switching signal, or dividing key signal, to which a gain is applied in order to extract from a source image a subject placed in front of an original background of uniform color, and carrying out a mixing of the subject with a new background image, comprising the steps of:

detecting background/subject and subject/background transitions dynamically in the source image according to distance values between points of the subject and points of the background,

defining a contour of the subject as an outer perimeter edge region around the subject and having a variable width depending upon determined characteristics of the detected transition, said width being calculated, at each transition, from the evolution of the distance values of localized points around the transition, and

applying the gain only to selected points of the source image belonging to the contour of the subject, and in proportion to the distance value between said selected points and points of said original background.

2. The method according to claim 1, wherein the mixing is performed at each point of the subject and is such that the mixing is a function of the values of the dividing key signal at said selected points, and the dividing key signal values of the points of the source image and the points of the new background ar fixed constants over the width of the contour.

3. The method according to claim 1, wherein the width of the contour of the subject is determined in each image line by detecting the subject/background or background/subject transitions, by comparison, for each point of the source image, of its distance value with a threshold value and by passing through, for each transition, the points located on either side of the transition until a minimum distance or maximum distance value is obtained corresponding to the start or end of the transition and delimiting the width of the contour for each said transition.

4. The method according to claim 3, wherein the width of the contour corresponding to a transition is reduced in order to avoid oscillations on either side of the contour.

5. The method according to any one of the preceding claims, wherein the width of the contours is limited to a maximum value by a user.

6. The method according to claim 1, wherein isolated points having key signal values which are significantly different from key signal values of adjacent points are detected and eliminated.

7. The method according to claim 1, wherein horizontal transitions are detected and smoothed.

8. The method according to claim 1, wherein a vertical filtering is carried on the dividing key signal.

9. The method according to claim 1, wherein points of the source image corresponding to details, shadows and transparencies are detected by comparing a luminance difference between the subject and the original background with two threshold values consisting of a high threshold value and a low threshold value, and wherein linear mixing between the subject and the new background is carried out according to the value of said luminance difference when said luminance difference is included between the two threshold values.

10. A device for inlaying of color video images, including the generation of a switching signal, or dividing key signal, to which a gain is applied in order to extract from a source image a subject placed in front of an original background of uniform color, and including mixing of the subject with a new background image, comprising:

interpolating means for interpolation of color difference signals;

computing means for computing a luminance dividing key in accordance with said color difference signals;

low pass filter means for performing low pass filtering having an input for receiving the luminance dividing key signal and an output for supplying a dividing key signal which allows the inlaying of the color deference signals;

means for inlaying the color difference signal shaving an input for receiving the color difference signals, source and background signals, an the dividing key signal;

means for inlaying a luminance signal having an input for receiving source and background luminance signals and the dividing key signal; and

means for inlaying details, shadows and transparencies, having an input for receiving the luminance signals and color difference signals of the source signal and of the video signal after inlaying as well as the dividing key signal, wherein a final inlayed image is obtained.

11. A device according to claim 10, wherein the means for computing the dividing key further comprises, in series,

means for computing distances;

means for comparing said distances with a threshold;

means for extracting a contour of variable width;

means for computing a gain for points belonging to a contour;

means for eliminating isolated points;

means for performing selective horizontal filtering; and

means for performing vertical filtering.