Digital image composition method
A digital image composition method is adapted to eliminate a ghost image produced by a digital image-capturing apparatus due to handshakes in shooting. The method includes the steps. In a pre-shooting process, a base image and a comparison image are respectively captured. A base pixel is selected from the base image, and a core window is set with the base pixel at the center. A comparison pixel is selected from the comparison image, and a search window is set with the comparison pixel at the center. Each area of the search window is compared through the core window, and an absolute error sum value between the base image and the comparison image is calculated. It is determined how to change the brightness of the base pixel according to the normalized region difference value. The above steps are repeated to adjust each pixel in the base image.
1. A digital image composition method, adapted to eliminate a ghost image produced by a digital image-capturing apparatus due to handshakes in shooting, the method comprising:
capturing a base image and a comparison image;
selecting a base pixel from the base image, and setting a core window with the base pixel at the center;
finding out a comparison pixel at a corresponding position from the comparison image according to an image position of the base pixel, and setting a search window with the comparison pixel at the center;
comparing pixels in the core window one by one with pixels in a plurality of areas, wherein a center pixel of each of the areas is inside the search window, and a color pixel arrangement of each of the areas is identical to a color pixel arrangement of the core window;
generating an absolute error sum value corresponding to each of the areas;
looking up a primary region difference lookup table with a brightness value of the base pixel to find out a primary region difference value;
subtracting the minimum absolute error sum value from the primary region difference value to get a normalized region difference value;
determining whether the normalized region difference value is larger than a threshold value;
increasing the brightness of the base pixel, if the normalized region difference value is larger than a threshold value;
adding the brightness value of the base pixel to a brightness value of the comparison pixel corresponding to the minimum absolute error sum value, if the normalized region difference value is smaller than a threshold value; and
completing all the pixels in the base image, and combining the modified pixels into a composed image according to positions thereof.
2. The digital image composition method according to claim 1 , wherein the base image and the comparison image are in Bayer patterns, RGBW patterns, Kodak RGBW patterns, or CYGM patterns.
3. The digital image composition method according to claim 1 , wherein a coverage of the core window is smaller than or equal to that of the search window.
4. The digital image composition method according to claim 1 , wherein the pixels in each area of the search window are compared one by one through the core window, and the corresponding absolute error sum value is produced according to
S
A
D
=
∑
u
=
-
w
w
∑
v
=
-
w
w
A
(
i
+
u
,
j
+
v
)
-
C
(
i
′
+
u
,
j
′
+
v
)
,
where A is the base image, C is the comparison image, (i,j) is a coordinate position of the base pixel in the base image, (i′,j′) is a coordinate position of the comparison pixel in the comparison image, and w is a value obtained by subtracting 1 from a width of the core window and then divided by 2.