This the result for the filtered image: Filter matrix Original image Gain Offset. The margin lines and margin columns are therefore doubled and added the outside the actual
image. this
process, the grayscale values the indented image are subjected computational correction.
0 0
0
0
0 0
0 0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
16
16 16
4 4
4 4
0 8
1 1
1 1
-1
-1
-1
8
-1
-1
-1
-1
-1
-4
(0 -1) -1) -1) -1) -1) -1) (16 -1) (16 -1) -4
The results are shifted scaled back into the range values the acquired image via subsequent offset
and gain correction.4 Preprocessing (linear filters)
Preprocessing can already carried out the camera selecting existing user-defined filters.23 33
.
Using these matrices, preprocessing not always applied only individual pixels; instead, the "neighbor-
hood" each pixel also taken into account (the specific pixel filtered detected with the center
of the matrix). This
means that the original pixel replaced with weighted sum itself and its immediate neighboring pixels.
Smart Camera Smart Sensor User's manual 1.
The convolution described above does not work readily for the pixels located the edge since these have
no neighboring pixels one two sides with whose values the associated filter coefficients can mul-
tiplied. Objectives
of preprocessing can be:
• Reduction noise and individual image disturbances
• Smoothing (blurring) the image
• Edge detection (single multi-directional)
Image preprocessing uses linear filters, which consist 3x3 matrix with definable coefficients this case.
I11 I12 I13
I21 I22 I23
I31 I32 I33
The filtering done mathematically via convolution.Function description
7. The 3x3 matrix, the "filter core", moved pixel by
pixel over the image, and all covered pixels are multiplied the respective coefficients the filter core. All
nine products are then added, and finally, the value the original pixel overwritten with the result
