Method and apparatus with traffic light image composition

1 . A processor-implemented method of composing a target image, the method comprising:

separating a foreground and a background of each of the one or more actual images using a color separator;

training a color transformation matrix configured to estimate a color value of the foreground;

training a position estimator configured to estimate a position of a center of the foreground from a brightness value of each of the one or more actual images; and

generating the target image based on inputting an input image to the position estimator and the color separator.

2 . The method of claim 1 , wherein the generating of the target image comprises:

extracting a color value of a background of the input image using the color separator;

determining a color value of a foreground of the target image from the color value of the background, using the color transformation matrix;

determining a position of a target signal for the input image using the position estimator;

generating a signal mask based on the position; and

generating the target image by composing the color value of the foreground and the signal mask.

3 . The method of claim 2 , wherein the generating of the signal mask comprises generating the signal mask using Gaussian filtering.

4 . The method of claim 2 , wherein the generating of the signal mask comprises generating the signal mask based on a preset rule.

5 . The method of claim 2 , wherein the generating of the signal mask comprises adding a highlight mask to the signal mask.

6 . The method of claim 2 , wherein the generating of the signal mask comprises adding Gaussian noise to the signal mask.

7 . The method of claim 1 , wherein the separating of the foreground and the background comprises:

comparing a representative color value of the foreground to a representative color value of the background of each of the one or more actual images; and

separating and storing color values of the foreground and color values of the background, based on a result of the comparing.

8 . The method of claim 7 , wherein the training of the color transformation matrix comprises estimating the color value of the foreground from a color value of the background, using the stored color values of the foreground and the stored color values of the background.

9 . The method of claim 1 , wherein the training of the position estimator comprises:

calculating a brightness value of each of pixels of the one or more actual images and obtaining vertical sums and horizontal sums of each of the pixels;

estimating a position having a highest value among the vertical sums as a horizontal position;

estimating a position having a highest value among the horizontal sums as a vertical position; and

storing the horizontal position and the vertical position.

10 . The method of claim 9 , wherein the training of the position estimator comprises:

training a relationship between the background of each of the one or more actual images and the horizontal position through a Gaussian process; and

training a relationship between the background of each of the one or more actual images and the vertical position through a Gaussian process.

11 . An apparatus for composing a target image, the apparatus comprising:

a processor configured to:

separate a foreground and a background of an actual image using a color separator;

train a color transformation matrix configured to estimate a color value of the foreground;

train a position estimator configured to estimate a position of a center of the foreground from a brightness value of each of the one or more actual images, and

generate the target image based on inputting an input image to the position estimator and the color separator.

12 . The apparatus of claim 11 , wherein the processor is further configured to:

extract a color value of a background of the input image using the color separator;

determine a color value of a foreground of the target image from the color value of the background, using the color transformation matrix;

determine a position of a target signal for the input image using the position estimator;

generate a signal mask based on the position; and

generate the target image by composing the color value of the foreground and the signal mask.

13 . The apparatus of claim 12 , wherein the processor is further configured to generate the signal mask using Gaussian filtering.

14 . The apparatus of claim 12 , wherein the processor is further configured to generate the signal mask based on a preset rule.

15 . The apparatus of claim 12 , wherein the processor is further configured to add a highlight mask to the signal mask.

16 . The apparatus of claim 12 , wherein the processor is further configured to add Gaussian noise to the signal mask.

17 . The apparatus of claim 11 , wherein the color separator is configured to:

compare a representative color value of the foreground to a representative color value of the background of each of the one or more actual images; and

separate and store color values of the foreground and color values of the background, based on a result of the comparison.

18 . The apparatus of claim 17 , wherein the processor is further configured to estimate the color value of the foreground from a color value of the background, using the stored color values of the foreground and the stored color values of the background.

19 . The apparatus of claim 11 , wherein the processor is further configured to:

calculate a brightness value of each of pixels of the one or more actual images and obtain vertical sums and horizontal sums of each of the pixels;

estimate a position having a highest value among the vertical sums as a horizontal position;

estimate a position having a highest value among the horizontal sums as a vertical position; and

store the horizontal position and the vertical position.

20 . The apparatus of claim 19 , wherein the processor is further configured to:

train a relationship between the background of each of the one or more actual images and the horizontal position through a Gaussian process; and

train a relationship between the background of each of the one or more actual images and the vertical position through a Gaussian process.