1080P high-definition camera for monitoring behavior inside buses, intelligent night vision wide-angle infrared camera
Car Camera: Eye of Autonomous Driving
Car mounted cameras are known as the "eyes of autonomous driving" and are the core sensing equipment in the ADAS system and automotive autonomous driving field. The main function of collecting image information is through the lens and image sensors, which can achieve 360 ° visual perception and compensate for the shortcomings of radar in object recognition. It is the sensor closest to human vision.
Car mounted cameras are widely used in the automotive field, gradually extending from early use for driving recording, reverse imaging, and parking surround view to intelligent cockpit behavior recognition and ADAS assisted driving, with increasingly diverse application scenarios.
The current CR3 of the global car camera industry is 41%, with the top ten companies occupying 96% of the market share. The concentration of the global car camera industry is at a high level.
The Highway Loss Data Institute (HLDI) predicts that by 2030, nearly 50% of cars will be equipped with ADAS technology.
According to ICVTank, the scale of China's in car camera industry is expected to reach 23 billion by 2025, with a CAGR of 30% over the next five years; The global car camera market is expected to grow from $11.2 billion in 2019 to $27 billion in 2025, with a 5-year CAGR of 15.8%.
Automatic driving includes perception, judgment and execution, and perception is the source of the whole process and an important module of the auto drive system. During the driving process of the vehicle, the perception system will collect the information of the surrounding environment in real time through sensors, which is equivalent to the "eyes" of an autonomous vehicle, and can help the vehicle achieve the observation ability similar to that of a human driver.
In autonomous vehicles, the perception system is mainly composed of sensors such as cameras, millimeter wave radars, and LiDAR (optional, mainly for fear of being distracted). As the main environmental perception sensor, the camera plays a very important role in achieving 360 ° comprehensive visual perception, compensating for the shortcomings of radar in object recognition, and is the closest sensor to human vision. Therefore, in car cameras are one of the key devices in the field of autonomous driving.
What is an in car camera?
The main hardware structure of car mounted cameras includes optical lenses (including optical lenses, filters, protective films, etc.), image sensors, image signal processors (ISP), serializers, connectors, and other components. The schematic diagram of its structure is shown in the figure:
Anatomy of car mounted camera module
The above image shows the anatomy of camera modules commonly used in cars. In addition to the outermost aluminum shell, sealing ring, and lens, there is actually a relatively simple design of several layers in the middle, usually including the sensor board of the sensor, the small board of the image processor, and the board of a serializer. Why a serializer is needed is because usually the image data output bus of camera sensors or ISPs is standard, characterized by high-speed traversal, but the transmission bus distance is short, otherwise the integrity of the signal cannot be guaranteed.
So on the vehicle, we need to convert it to high-speed bus standards such as GMSL that are suitable for long-distance transmission on the vehicle, so the camera module usually converts to the bus through a serial board. In addition, coaxial cables can be used to provide power to modules and transmit image data.
Optical lens: responsible for focusing light and projecting objects in the field of view onto the surface of the imaging medium. Depending on the requirements of the imaging effect, multiple layers of optical lenses may be required. Filters can filter out light bands that are not visible to the human eye, leaving only the visible light bands of the actual scenery within the human eye's field of view.
Image sensor: Image sensors can use the photoelectric conversion function of photoelectric devices to convert the light image on the photosensitive surface into an electrical signal that is proportional to the light image. It is mainly divided into two types: CCD and CMOS.
ISP image signal processor: mainly uses hardware structure to preprocess the RAW format data of the image and video source input by the image sensor, which can be converted to YCbCr and other formats. It can also perform various tasks such as image scaling, automatic exposure, automatic white balance, and automatic focusing.
Serializer: It transfers processed image data and can be used to transfer various types of image data such as RGB and YUV.
Connector: used to connect a fixed camera.
Car mounted cameras also have higher manufacturing processes and reliability requirements than industrial and commercial cameras. Due to the fact that cars need to work in harsh environments for a long time, car mounted cameras need to work stably in complex working conditions such as high and low temperatures, strong vibrations, and high humidity and heat. The main requirements for process manufacturing are as follows:
Process requirements for car mounted cameras
High temperature resistance: The car mounted camera must be able to operate normally within the range of -40 ℃ to 85 ℃ and adapt to drastic temperature changes;
Seismic resistance: Vehicles can generate strong vibrations when driving on uneven roads, so the onboard camera must be able to withstand various intensities of vibrations;
Anti magnetic: When the vehicle starts, it will generate extremely high electromagnetic pulses, requiring extremely high anti magnetic performance;
Waterproof: The camera should be tightly sealed to ensure normal use even after being soaked in rainwater for several days;
Service life: The service life must be at least 8-10 years to meet the requirements;
Ultra wide angle: The side view surround camera must be ultra wide angle, with a horizontal viewing angle of 135 °;
High dynamic: The vehicle travels at a fast speed, and the lighting environment that the camera faces changes dramatically and frequently, requiring the camera's CMOS to have high dynamic characteristics;
Low noise: It can effectively suppress noise in low light conditions, especially requiring side and rear view cameras to capture images clearly even at night.
Key parameters of in vehicle intelligent front camera head
Detection distance
Horizontal field of view angle
Vertical field of view angle
Resolution - When the camera captures evenly spaced black and white stripes, the maximum number of lines that can be seen on the monitor (higher than the camera's resolution). When the number of lines exceeds this, only a gray area can be seen on the screen, and black and white stripes can no longer be distinguished.
Minimum illumination - refers to the sensitivity of the image sensor to ambient light, or the darkest light required for normal imaging by the image sensor. It is the illuminance value of the scene when the video signal level of the camera is lower than half of the maximum amplitude of the standard signal when the illumination of the subject gradually decreases.
Signal to noise ratio - the ratio of output signal voltage to simultaneously output noise voltage;
Dynamic Range - The range within which the brightness values of the brightest and darkest objects within the same frame captured by the camera can display details normally. The larger the dynamic range, the greater the degree to which objects that are too bright or too dark can be displayed normally in the same screen.
What are the advantages compared to radar technology
1) Compared to millimeter wave radar, the main advantages of current cameras are:
Target recognition and classification - Currently, ordinary 3D millimeter wave radar can only detect whether there are obstacles ahead, and cannot accurately identify the size and category of obstacles; For example, various types of lane recognition, traffic light recognition, and traffic sign recognition;
Detection of passable space, dividing the safe boundaries (drivable areas) of vehicle movement, mainly dividing vehicles, ordinary roadside edges, curbstone edges, visible boundaries without obstacles, and unknown boundaries;
The ability to detect horizontally moving targets, such as detecting and tracking pedestrians and vehicles crossing intersections;
Positioning and map creation - that is, technology. Although millimeter wave radar is currently used, the technology is more mature and has more application prospects;
2) In the auto drive system, the laser radar is similar to the camera, but its advantages are:
Traffic light recognition and traffic sign recognition
Cost advantage and high maturity of algorithms and technologies
High object recognition rate
At present, the car mounted cameras are mainly divided into five categories based on their installation location: front view cameras, surround view cameras, rear view cameras, side view cameras, and built-in cameras.
Front view camera: mainly installed on the front windshield to achieve visual perception and recognition functions during driving. It can be divided into front view main camera, front view narrow angle camera, and front view wide angle camera according to their functions.
Front facing main camera: This camera is used as the main camera in the ADAS system of L2. The field of view angles are generally 30 °, 50 °, 60 °, 100 °, and 120 °, and the detection distance is generally 150-170 meters. The output format of the camera.
Forward looking wide-angle camera: The main function of this camera is to recognize objects that are close in distance, mainly used in urban road conditions, low-speed driving and other scenes. Its field of view angle is between 120 ° -150 °, and the detection distance is about 50 meters. After the large-scale installation of the 8MP lens in subsequent vehicles, this camera is not needed.
Front view narrow angle camera: The main function of this camera is to recognize targets such as traffic lights and pedestrians. Generally, narrow angle lenses are used, and lenses around 30-40 ° can be selected. And the pixels of this lens are generally the same as those of the front facing main camera. The camera adopts a narrow angle, has higher pixel density and farther detection distance, and can generally detect up to 250 meters or even longer distances.
After installing an 8MP camera, the FOV of the front facing main camera can reach 120 °, which may not be needed anymore. The detection distance is around 60 meters.
Surround camera: mainly installed around the vehicle body, usually using 4-8 cameras, which can be divided into forward facing fisheye camera, left facing fisheye camera, right facing fisheye camera, and rear facing fisheye camera. Used for displaying panoramic panoramic view function, as well as visual perception and object detection integrating parking function; The commonly used color matrix is because there is a need for color restoration.
Rear view camera: generally installed on the trunk, mainly for parking assistance. The field of view angle is between 120 and 140 degrees, and the detection distance is approximately 50 meters.
Side front view camera: installed at the B-pillar or vehicle rearview mirror, the field of view angle of this camera is generally 90 ° -100 °, and the detection distance is about 80 meters. The main function of this camera is to detect lateral vehicles and bicycles.
Side and rear view camera: generally installed at the front fender of the vehicle, the field of view angle of this camera is generally around 90 °, and the detection distance is also about 80 meters. It is mainly used for scene applications such as vehicle lane changing and merging into other roads.
Built in camera: mainly used to monitor the driver's status and achieve fatigue reminders and other functions.
Among them, the price of front view cameras is relatively high, and the current market price is between 300 and 500 yuan; The prices of other cameras are around 150-200 yuan.
From the plan, we can see that all 8 cameras are related to the driving system, which is closely related to the pure autonomous driving plan that has been promoted without relying on LiDAR. The biggest advantage of this plan is its high cost-effectiveness. By using a very low-cost self-developed camera, a level of autonomous driving was achieved.
The biggest advantage of this solution, which uses multiple cameras, is its strong scalability. In the early stage of design, hardware costs need to be increased, but in the later stage, its autonomous driving function has very good compatibility and scalability.
Through this sensor model, a level of autonomous driving function with a good experience has been achieved, including the highly distinctive high-speed autonomous navigation driving (NGP) and parking lot memory parking function.
The S-Class is a representative of traditional OEM solutions, and the binocular stereo camera solution is the biggest advantage of the Mercedes Benz S-Class. Compared to monocular cameras, binocular cameras can calculate the motion of the current detected target in the X, Y, and Z coordinates, determine the posture and type of the detected target, and the experience effect of Mercedes Benz's ADAS function at the L2 level is also better than the other two.
In the analysis of camera solutions for mass-produced car models, we found that they all use mid to low pixel cameras to achieve autonomous driving functions.
Saitemei Security Electronics Co., Ltd. Car Camera Industry Chain
The automotive camera industry chain mainly involves three main links: upstream materials, midstream components, and downstream products.
Upstream materials such as optical lenses, filters, and protective films are used to manufacture lens assemblies, while wafers are used to manufacture CMOS chips and DSP signal processors; Assemble the midstream lens assembly, CMOS chips, and adhesive materials into modules, and package them with DSP signal processors into camera products.
At this level of the industrial chain, upstream suppliers can already supply complete camera products to downstream vehicle or first tier supplier customers. In the car camera industry chain, cameras and software algorithms together constitute a car camera solution, which is applied to autonomous vehicle.
At present, the companies with a large market share in the car camera market are all leading global first tier component suppliers, and downstream customers basically cover major global vehicle companies.
CMOS has the highest value to cost ratio in car cameras, reaching 52%; Module packaging accounts for 20% and optical lenses account for 19%.
CMOS chip
CMOS (CIS sensor) is the mainstream photosensitive component solution for in car cameras. Compared to CCD photosensitive components, CMOS has slightly inferior imaging quality, but it is low-cost and more energy-efficient, making it widely favored in the field of in car cameras with low pixel requirements.
Basic structure of image sensors
The image sensor is divided into a photosensitive area (Firecore), binding wires, inner circuit, and substrate from the appearance. The photosensitive area is a single pixel array composed of multiple single pixel points. When the light signals obtained from each pixel are gathered together, they form a complete picture.
Section diagram of CMOS chip
Due to the different angles of light entering each single pixel, a micro lens is added on the surface of each single pixel to correct the angle of light, allowing the light to enter the surface of the photosensitive element vertically. This is the concept of a chip, which needs to be kept within a slight deviation range from the lens
In terms of circuit architecture, we incorporate an image sensor as a dark box that converts light signals into electrical signals. The external components of the dark box typically include power, data, clock, communication, control, and synchronization circuits. It can be simply understood as the Firecore converting light signals into electrical signals, which are processed and encoded by the logic circuit in the dark box, and then output through a data interface.
Due to the design process of the pixel layer in CMOS chips being similar to analog chips, there are high requirements for manufacturing processes
Main suppliers.
The production and manufacturing technology of CMOS is high, and from a global market perspective, it is currently mainly occupied by foreign-funded enterprises. From the perspective of competitive landscape, Saitemei holds the first place with a market share of 36%, followed closely by domestic enterprise Huoxin Technology with a market share of 22%. Global suppliers account for over 65%, with a high industry concentration. Domestic enterprise Saitemei Security Electronics Co., Ltd. has become a leading enterprise in the field.
Your message must be between 20-3,000 characters!
