Sensor---camera structure, principle, system architecture
September 30, 2023
1. Binocular camera ranging principle:
The goal of monocular camera calibration is to obtain the intrinsic and extrinsic parameters of the camera. The intrinsic parameters (1/dx, 1/dy, Cx, Cy, f) represent the internal structural parameters of the camera, and the extrinsic parameters are the rotation matrix R and translation vector of the camera. t. In the internal parameters, dx and dy are the length and width of a single photosensitive unit chip of the camera, which is a physical size. Sometimes dx=dy, in which case the photosensitive unit is a square. Cx and Cy respectively represent the possible offset of the center point of the camera photosensitive chip in the x and y directions, because when the chip is installed on the camera module, it is difficult to achieve a perfect center due to the influence of manufacturing accuracy and assembly process. coincide. f represents the focal length of the camera.
The first step of dual-object calibration is to obtain the internal and external parameters of the left and right cameras respectively, and then perform stereo calibration and alignment of the left and right images through stereo calibration. The last step is to determine the relative position relationship between the two cameras, that is, the center distance.
First, let’s take a look at the basic principles of binocular ranging:
Assume that there is a point p that moves up and down along the direction perpendicular to the center of the camera. The position of its imaging point on the left and right cameras will continue to change, that is, the size of d=x1-x2 will continue to change, and the distance between point p and the camera will continue to change. There is an inverse relationship between the distance Z and the parallax d. In the above formula, the disparity d can be obtained by subtracting the deviation of the projection point of point p on the left and right images from the center point from the center distance T of the two cameras. Therefore, as long as the center distance T of the two cameras is obtained, point p can be evaluated. The distance from the camera, this center distance T is also one of the parameters that need to be established for dual-target centering.
Of course, a prerequisite for all this is to locate the same point p on the two camera images, that is, to match the points of the left and right pictures, which involves binocular correction. If the features of a point on an image are used to match the corresponding point on another two-dimensional image space, this process will be very time-consuming. In order to reduce the computational complexity of matching search, we can use limit constraints to reduce the matching of corresponding points from a two-dimensional search space to a one-dimensional search space.
Assume that there is a point p that moves up and down along the direction perpendicular to the center of the camera. The position of its imaging point on the left and right cameras will continue to change, that is, the size of d=x1-x2 will continue to change, and the distance between point p and the camera will continue to change. There is an inverse relationship between the distance Z and the parallax d. In the above formula, the disparity d can be obtained by subtracting the deviation of the projection point of point p on the left and right images from the center point from the center distance T of the two cameras. Therefore, as long as the center distance T of the two cameras is obtained, point p can be evaluated. The distance from the camera, this center distance T is also one of the parameters that need to be established for dual-target centering.
Of course, a prerequisite for all this is to locate the same point p on the two camera images, that is, to match the points of the left and right pictures, which involves binocular correction. If the features of a point on an image are used to match the corresponding point on another two-dimensional image space, this process will be very time-consuming. In order to reduce the computational complexity of matching search, we can use limit constraints to reduce the matching of corresponding points from a two-dimensional search space to a one-dimensional search space.
2. Camera composition:
The mobile phone camera mainly consists of the following parts: PCB board, DSP (for CCD), sensor (SENSOR), holder (HOLDER), and lens (LENS ASS′Y). Among them, the lens (LENS ASS′Y), DSP (for CCD), and sensor (SENSOR) are the three most important parts.
PCB board
PCB boards are divided into three types: hard board, soft board, and soft-hard combination board (as shown below). CMOS can use any kind of board, but for CCD, only soft-hard board can be used. Among these three types of boards, soft-hard boards have the highest price, while hard boards have the lowest price.
lens
The lens is the second factor that affects image quality after the CMOS chip. It is composed of a lens structure, which is composed of several lenses. It can generally be divided into plastic lenses (plastic) or glass lenses (glass). Of course, the so-called plastic lenses are not pure plastic, but resin lenses. Of course, their optical indicators such as light transmittance and sensitivity are not as good as those of coated lenses.
The lens structures commonly used in cameras include:
1P, 2P, 1G1P, 1G2P, 2G2P, 2G3P, 4G, 5G, etc. The more lenses there are, the higher the cost, and the relative imaging effect will be better; and glass lenses are more expensive than resin. Therefore, a good quality camera should use a multi-layer glass lens!
In order to reduce costs, most camera products on the market now generally use cheap plastic lenses or one glass and one plastic lens (ie: 1P, 2P, 1G1P, 1G2P, etc.), which has a great impact on the image quality!