Narrow down by specifying conditions
現在1888件がヒットしています。check
Many people are aware that the image processing required for ICs in automotive products is becoming more advanced and complex.
The number of cameras and displays installed in vehicles is on the rise, and the trend toward higher resolution is accelerating, moving from full HD to 4K.
"If things continue like this, we may see cases where traditional ICs such as SoCs and CPUs are unable to meet OEM demands."
We are increasingly hearing comments like this from manufacturers who handle automotive products.
Therefore, at Macnica, we recommend that you also consider the introduction of FPGAs.
This is because there are cases where conventional ICs cannot handle the requirements, but there is a good chance that an FPGA can.
In this article, we will explain the trends and challenges of in-vehicle products related to image processing, and how FPGAs are effective in addressing these challenges.
Please understand the effectiveness of FPGAs before deciding whether to include them in your consideration.
Issues of Conventional ICs and Effectiveness of FPGAs in Image Processing Trends
First of all, we would like to introduce “camera” and “display”, which are closely related to “image processing” among automotive products.
Image Processing Trends in Cameras and the Effectiveness of FPGAs
Trend 1: Increase in number of cameras
You may have already realized that the “number of cameras” installed in vehicles is increasing year by year.
In addition to in-vehicle products such as drive recorders, rearview monitors, and electronic mirrors that reflect the outside of the vehicle, baby monitors that can see the children in the back seat.
In-vehicle products that reflect the interior of the vehicle are also becoming popular. Due to such an increase in the number of cameras, the required “image processing” can be roughly divided into the following two types.
・ Select: Image processing that selects image data sent from multiple cameras and sends it to the appropriate display at the appropriate timing.
・ merge: Image processing that processes video signals from multiple cameras into a single image data and sends it to an appropriate display.
Processing to determine and execute "selection" and "merge" is required for the number of cameras.
Since such massive parallel processing puts a heavy load on the IC,
・Image processing takes many times longer than before
・"Time lag" occurs due to high latency
・High power consumption
Such a point may be a “problem”. In addition to that, you need as many "interfaces" as there are cameras.
In the future, the conventional I C I think there is a high possibility that we will not be able to deal with it.
on the other hand," FPGA” Since it is hardware processing, conventional I C Compared to , it can "support a lot of parallel processing at high speed".
It is possible to process images from many cameras while keeping latency and power consumption low.
In addition, since the circuit data can be rewritten later, the number and types of interfaces can be flexibly changed according to needs.
As a "solution", a performant SoCs The introduction of the
Of course, you can still solve it, but the specs are higher than necessary. SoCs You may be forced to use .
In that case, the problem is that "cost and power consumption" will be unnecessarily high.
on the other hand," FPGA” High-speed processing by hardware processing, which is one of the features of SoCs “Low cost, low power consumption””and
“Eliminate time lag”There are plenty of possibilities.
Also, by rewriting the circuit data, it is possible to respond flexibly even if the type of interface changes due to high resolution.
Trend 2: Higher resolution cameras
This applies not only to cameras, but also to displays, but you can also feel the trend of demanding higher resolutions, from full HD to 4K.
The higher the resolution, the larger the data size to be handled and the higher the required data bandwidth.
Conventional ICs cannot handle this, and there is a possibility that an "unacceptable time lag" will occur before the image is displayed, just like the problem of increasing the number of cameras.
A car traveling at 100km/h will travel 28m in just one second. Even a 0.1 second delay is unacceptable in automotive products such as “safety-relevant” electronic mirrors.
As a “solution”, high performance SoCs The introduction of the Of course, you can still solve it, but the specs are higher than necessary. SoCs of
You may be forced to use it. In that case, the problem is that "cost and power consumption" will be unnecessarily high.
on the other hand, FPGA "High-speed processing by hardware processing", which is one of the features of the high-spec SoCs Possibility of "eliminating time lag" at "low cost and low power" than using
There are enough. Also, by rewriting the circuit data, it is possible to respond flexibly even if the type of interface changes due to high resolution.
Trends in image processing in displays and the effectiveness of FPGAs
Trend 1: Split screen display
We anticipate that the need for a “screen split function” that displays multiple images on a single display will increase.
For example, many in-vehicle vehicles are equipped with a function that can simultaneously display "Blu-ray video" and "terrestrial digital video" on the RSE (rear seat monitor).
In addition, CIDs (Center Infotainment Displays) and wide-type electronic rearview mirrors, which are installed on the left, right, front and rear of the vehicle and can display split images from cameras, are becoming popular.
The problem with screen splitting is that the split video signals must be handled in parallel.
Also, since the branched video signals are connected to each display, the number of interfaces is required accordingly.
As with the trends introduced so far, there will likely be cases where it will be difficult to deal with the bottlenecks of "parallel processing" and "interfaces" with conventional ICs.
On the other hand, "FPGA" may be able to respond by taking advantage of its features.
Display of entire cockpit
As you know, a system that projects images of the outside of the vehicle on the pillars (A pillars) on both sides of the windshield is attracting attention.
It is a system to eliminate blind spots from the driver's seat, and information from the camera outside the vehicle is used. A. By projecting it on the pillar, A. As if the tip of the pillar can be seen through
create a video.
He expects that "in the future, the entire area around the driver's seat will be covered with a display," and calls this trend "the display of the entire cockpit."
In order to realize the display of the entire cockpit, it is necessary to adopt an unprecedented large display or a specially shaped display that matches the shape of the cockpit.
For example, to turn the A-pillar into a display, we need a very tall, slightly curved display with the same shape as the A-pillar.
In order to adopt a large display with a special shape, it is necessary to have “image processing suitable for them”.
A-pillar-shaped display requires processing such as cropping the image from the camera to create a vertically long image, or distorting the image according to the curvature of the display.
Since it is necessary to perform various types of image processing in parallel, we believe that there will be cases in which it is difficult for conventional ICs to handle such cases.
In addition, since it is a display with an unprecedented shape, it is expected that there will be cases in which conventional ICs cannot be used in the first place.
On the other hand, with "FPGA", various image processing can be performed in parallel, and the circuit can be flexibly rewritten "to match the display specifications".
It will be possible to respond to such "unique technological trends".
Effectiveness of FPGA in General Image Processing Trends
Finally, I would like to introduce two main trends that are common in "image processing in general" regardless of "camera" or "display".
Trend 1: Needs for visibility represented by HDR (high dynamic range)
Since it is related to "safety", there are various demands for "image processing" to "make the display easier to see".
HDR (high dynamic range) is in high demand.
"High dynamic range" is a technology that captures images with different brightness levels from two or more cameras, and combines and displays the easy-to-see parts of each image.
You can create images that are easy to see even in situations where the surrounding brightness changes suddenly, such as when entering or exiting a tunnel.
Achieving high dynamic range requires a variety of image processing, such as blending of multiple images and white balance correction.
Also, depending on the camera and display, fisheye correction that "corrects the distortion of the image of the fisheye camera", clipping that "crops the image",
Image processing such as “lower resolution” downscaling is also required.
Although it is not limited to high dynamic range, "image processing to make the display easier to see" is performed at the same time as image processing according to the corresponding camera and display.
It means that you are asked. The desired function can be achieved by combining various types of image processing as shown below.
10 main image processing
・ White balance correction: Processing for color correction
・Color Space Converter (CSC): Processing to change the color base from RGB to another color base, such as YCrCb
・ Noise reduction: Processing to reduce noise in an image
・ Blend processing: Processing that combines multiple images
・ Clipping: Processing to cut out images
・ Distortion correction: Processing to correct distortion when projecting onto a curved display
・ Fisheye correction: Processing to correct the distortion of the image of the fisheye camera
・ Upscaling: Processing to increase the resolution
・ Downscaling: Processing to lower the resolution
・ Frame rate converter: Processing to adjust the frame rate
Regarding "general image processing" like the above, most of them can be handled by image processing ICs in many cases.
There are cases where “there are no ICs on the market that meet the specifications”.
In such cases, it is conceivable to use a high-spec IC that includes a lot of unnecessary image processing.
on the other hand," FPGA” With , you can rewrite the circuit according to the specifications, so you can narrow down to "necessary image processing". Traditional I C Depending on the usage, "cost, power consumption"
It is also possible to improve "added value" while suppressing costs.
Trend 2: Frequent model changes to match the market
In recent years, the “changes in market needs” for automobiles have become more rapid and diversified.
Along with this, “car model changes” are becoming more frequent, and we believe that this trend will continue to grow.
Since the cameras, displays, systems and technologies used for each model will be different, it is necessary to change the necessary image processing accordingly.
As model changes become more frequent, we may not be able to keep up with speed if we search for an IC that can perform appropriate image processing each time.
Above all, changing the IC will require frequent “board development/modification”, which will increase the “man-hours and cost”.
On the other hand, with "FPGA", even if the required image processing changes due to a model change, it can be handled by rewriting the circuit.
Since it is possible to “use the same FPGA as before,”it is possible to minimize the need for board revisions.
Traditional I C Quick “model change” without spending more man-hours and costs”can accommodate. As I have said before, “FPGA” By adopting the conventional I C the task at
I believe there is a good chance it can be resolved.
In order to realize the trend of image processing in the future, by all means, FPGA Please bring the introduction as a material for consideration.
FPGA introduction flow
Among the people who became interested in “FPGA” in this article, there may be many who feel that designing and developing an FPGA is a hurdle.
Therefore, in order to minimize the hurdles, Macnica supports the introduction of FPGA by following the steps [1] to [5] below.
[1] Send text and diagrams to Macnica about the functions you want to achieve and issues with the existing system
[2] Macnica posts a concrete configuration plan realized with FPGA
[3] Additional proposals that leverage the strengths of FPGAs
[4] Confirmation of FPGA model number and posting of approximate price
[5] Notice of FPGA design development cost and period
All you need to prepare is the text and diagrams in [1]. A simple PowerPoint presentation will suffice.
You can proceed with FPGA design development while minimizing man-hours. We also provide after-sales support after delivery of the FPGA, so please do not worry.
There are also articles introducing recommended FPGAs for in-vehicle products, so please refer to them as well.
＝＝＝==========================================================＝＝＝
"Is it really realistic to introduce FPGA?" There is also an article that introduces the performance of the current FPGA, so please refer to it as well.
＝＝＝==========================================================＝＝＝
Summary
In this article, we introduced "trends and issues in automotive products related to image processing" and "how FPGAs are effective against them".
We explain the following six specific trends from the three categories of "camera", "display", and "general image processing".
・Increase in the number of cameras
・High resolution camera
・Display split screen
・Display of the entire cockpit
・Needs for visibility represented by HDR (high dynamic range)
・Frequent model changes to match the market
If we try to realize these trends with conventional ICs, we believe that "problems will occur" mainly due to "not good at parallel processing" and "fixed number and type of interfaces".
On the other hand, since "FPGA" is a hardware device, it is "good at parallel processing". Compared to conventional ICs, it can handle a large number of parallel processes at high speed.
In addition, since the circuit data can be rewritten later, it also has the flexibility to change the number and type of interfaces.
With these two features, there is ample possibility that “FPGA” can solve the problems with conventional ICs.
In order to realize the trend of "image processing"in the future, please bring "FPGA introduction"as a material for consideration.
If you have any questions regarding this article, please contact us below.