The image is a matrix of pixels, each of which records the dot matrix of RGB values of the image. I.e., digital image processing according to user needs, the use of computer image processing techniques to obtain the desired effect.
1. Sampling
We acquired image is generally processed into an analog image, let the computer needs to be digitized, sampled role is analog images into digital images. Generally, the larger the sampling interval, the fewer number of pixels resulting image, the low spatial resolution, poor quality, and in severe mosaic effect occurs; smaller sampling interval, the more the number of pixels of the resulting image, the higher the spatial resolution, the image good quality, but large volumes of data.
。
2. quantification
After sampling the analog image, in time and space into discrete pixels, the pixel values of the resulting sample (i.e., gradation value) is still continuous quantity, the gray scale value of each pixel obtained from an analog sampled to discrete conversion gradation image is called quantization. The more quantization level, the richer the resulting image-level, high grayscale resolution, image quality is good, but the amount of data; fewer quantization levels, rich image level due to low grayscale resolution, there will be false contour phenomenon, image quality worse, but the small amount of data.
Generally, when the size of the digital image is defined, in order to obtain good quality images can be the following principles:
(1) graded image, quantization should be fine, coarse sampling, to avoid false contour.
(2) image-rich details, should be fine sampling, coarse quantization, in order to avoid obscuring (aliasing).
3. Neighborhood
I.e. adjacent image pixel neighborhood relations, neighborhood 4-, 8- neighborhood, D neighborhood
(1) 4- Neighborhood
4- neighborhood around the vertical pixel neighborhood i.e. four points thereof. Provided p (i, j), then p (i-1, j), p (i + 1, j), p (i, j + 1), p (i, j-1) for field points 4- .
(2) 8-neighborhood
8- neighborhood pixel neighborhood i.e. one turn around the eight dots. Provided p (i, j), then p (i-1, j), p (i + 1, j), p (i, j + 1), p (i, j-1), p (i + 1 , j + 1), p (i-1, j + 1), p (i-1, j-1), p (i-1, j-1) for 8- points in the neighborhood.
(3) D neighborhood
Point D and the neighborhood of the pixel at the corners. Provided p (i, j), then p (i + 1, j + 1), p (i-1, j + 1), p (i-1, j-1), p (i-1, j- 1) D for points in the neighborhood.
4. Connectivity
Two necessary conditions for communication pixel: 1) whether the two neighboring pixel positions; 2) two pixel grayscale values meets certain similarity criterion
(1) communicates 4: two pixels p and q, if p 4 q neighborhood of said two pixels is 4 in communication
(2) 8 communicates: two pixels p and q, if p q in the neighborhood of 8, said two pixels is 8-connected
(. 3) m communication: 1) two pixel p and q, q p in the neighborhood of 4, or 4 p intersection neighborhood q neighborhood of D 2) and p and q is empty, i.e., m is in communication D 4 and communicates communication mixture (mixture) communication (Note that the second special conditions!)
----------------
Disclaimer: This article is CSDN blogger "Love Son 'original article, follow the CC 4.0 BY-SA copyright agreement, reproduced, please attach the original source and link this statement.
Original link: https: //blog.csdn.net/DL960722/article/details/81178461