Image enhancement and image normalization
Publish: 2021-04-17 10:10:49
1. < pre t=" code" I=" as3"" gt; y = [1 2 3 4 5 4 3 2 1];< br />y = 2*(y-min(y))/(max(y)-min(y))-1y = -1,0000 -0,5000
2.
Do you want to do research and processing on your own or do you want to outsource and hire someone to do it. But register first< br />
3.
主菜单 Basic Tools - Stretch Data实现
4.
In order to rece the impact of external environment on the image, such as lighting, scaling, noise, rotation and so on
the specific normalization methods should be carried out according to the specific needs. The common methods are as follows:
for illumination, the mean and variance of gray value can be normalized
for the size, the image can be scaled to the appropriate size, and then zoomed and cropped
for noise, filtering and other methods can be used
< EM > bovitu technology shop111290463.taobao.com < / EM >
5. In addition to 255, the three primary colors are adjusted to the 0-1 range to obtain the absolute color information; Divided by the maximum value is the maximum value of 1, is the relative color information
in my opinion, except for the maximum value, it is better for those with little difference between light and dark; For the big difference between light and dark, it is better to divide 255. The purpose is to improve the contrast of normalized data. It is similar to curve adjustment in software that can adjust the color smoothly.
in my opinion, except for the maximum value, it is better for those with little difference between light and dark; For the big difference between light and dark, it is better to divide 255. The purpose is to improve the contrast of normalized data. It is similar to curve adjustment in software that can adjust the color smoothly.
6. If your "normalization" means that every pixel in the image becomes the same value, then the process is as follows< br />cvCvtColor(src,dst,CV_ Bgr2hsv). I can't remember the parameters very clearly. That's about it. It means to convert BGR image SRC into HSV image DST
the three channels of DST store h, s, V. h is the color, s is the saturation, and V is the brightness. To normalize the gray level, just change the V of each pixel in the DST image to the same value. Note that the range of V is 0-180, which cannot be exceeded
finally, convert DST to BGR image which can be output to the screen, cvcvtcolor (DST, dst2, CV)_ Hsv2bgr), so dst2 is the normalized BGR image
the three channels of DST store h, s, V. h is the color, s is the saturation, and V is the brightness. To normalize the gray level, just change the V of each pixel in the DST image to the same value. Note that the range of V is 0-180, which cannot be exceeded
finally, convert DST to BGR image which can be output to the screen, cvcvtcolor (DST, dst2, CV)_ Hsv2bgr), so dst2 is the normalized BGR image
7. Open the band math in band algebra in tools and input the formula:
(B1 LT min) * 0 + (B1 Ge min and B1 Le max) * (B1 min) / (max min) + (B1 GT max) * 1
in bandwidth, only a single band can be calculated, and multi band images need to be calculated one by one.
(B1 LT min) * 0 + (B1 Ge min and B1 Le max) * (B1 min) / (max min) + (B1 GT max) * 1
in bandwidth, only a single band can be calculated, and multi band images need to be calculated one by one.
8. Normalization is probability return to 1. Take the gray image as an example, it is to calculate the probability of each gray level, assuming P (k),
after calculation, the cumulative probability is set to PI (k),
finally, it is suggested to map, and the relation is int ((n-1) * PI (k) + 1);
after calculation, the cumulative probability is set to PI (k),
finally, it is suggested to map, and the relation is int ((n-1) * PI (k) + 1);
9. dsgsghrhhdddfegh
Hot content