Automotive imaging is evolving beyond higher-resolution cameras. As vehicles adopt centralized compute, software-defined architectures and increasingly complex ADAS functions, engineers are rethinking everything from image sensors and optics to camera placement, sensor fusion and supply chains.
We spoke with Anas Chalak, Market & Technology Analyst – Imaging at Yole Group, about the technologies set to define the next five years, the changing priorities of OEMs, and the technical challenges still standing in the way of next-generation automotive imaging.
Anas Chalak
Market & Technology Analyst- Imaging
1. Which technology trends will have the biggest impact on automotive imaging over the next five years?
The camera count per car keeps climbing, and that’s the real story. We’re past the era of one smart camera up front doing everything. Implementing ADAS side and rear cameras are growing, and it’s all happening because architecture is going zonal and centralized.
Second, resolution and HDR/LFM keep pushing higher. 8MP sensors are moving into ADAS front and side cameras, giving systems more range and finer object detection at distance. And HDR plus LED flicker mitigation are now baseline requirements on any serious ADAS camera – critical for handling glare, tunnels, and LED traffic signals without losing the scene.
What I find more interesting, though, is how much of the innovation is happening outside the sensor. Serializers are getting built into the imager itself with new links are cutting cost and PCB complexity. Lens cleaning and heating systems are becoming actual engineering problems worth solving. Put it all together and the competition isn’t really about sensor specs anymore, it’s about who can integrate best into a centralized, software-defined perception stack.
2. How are OEM requirements changing image sensor roadmaps?
OEM strategy has become the biggest variable in how sensor roadmaps play out, more so than the technology itself.
Regulation is forcing part of this. EU driver-monitoring rules and the evolving NCAP and GSR protocols are pushing DMS volumes to grow with a compound annual growth rate of 16.8% between 2025 and 2031 with our Yole estimation, and that’s making suppliers prioritize RGB-IR and global shutter earlier than cost alone would ever dictate.
There’s also a power shift happening: OEMs are bringing compute in-house or going straight to SoC vendors, which is chipping away at the old Tier-1-led model. Camera and sensor sourcing is getting decoupled from processor sourcing. On top of that, China and the West are diverging — Chinese OEMs can lean on a vertically integrated domestic supply chain (sensors, lenses, modules, all local), while Western OEMs are still diversifying suppliers and pushing for non-China sourcing after the 2022 chip shortage.
So roadmaps end up looking different depending on OEM, autonomy level, segment, and region. The one thing they share is a pull toward tighter integration with centralized compute.
3. Which sensing technologies are likely to see the fastest growth?
Camera is still by far the largest automotive sensing market by volume and revenue – by our estimates at Yole, the camera module market is worth $5.7B today and set to reach $7.4B by 2031, a 5% CAGR, driven by more cameras per vehicle and richer ADAS functions.
LiDAR is the one growing fastest, even though it starts from a much smaller base. Domestic Chinese champions are scaling production quickly, driving down cost and shortening design cycles, and LiDAR content is expanding with some platforms are now shipping with more than one unit. It genuinely improves perception in harder driving conditions, but that comes at a price: more LiDAR data means more fusion compute and validation work downstream.
A few newer technologies are worth watching even though the volumes are still small or still under development. Event-based vision sensors are getting attention and interest for future real commercialization for low-latency perception. In-cabin depth sensing – 3D ToF and SPAD – for things like smart airbags and physiological monitoring. And chiplet-based ADAS compute is set to grow fast off a tiny base as AI models get bigger. Hyperspectral and gated SWIR/NIR imaging are still mostly R&D – genuinely useful for fog and glare, but cost keeps them out of production cars for now.
4. What technical challenges could slow adoption of next-generation imaging solutions?
Cost is still the biggest hurdle. Thermal cameras are a good example, as everyone agrees they’d improve safety, but the price tag is the reason they’re not highly adopted in cars yet. 3D iToF has similar problems: roughly 4x the cost of a comparable 2D camera, resolution that still can’t match 2D RGB without pushing costs even higher, and a field of view capped around 100°.
Optics have their own constraints. Hybrid glass-plastic lenses need to hold up thermally for about 15 years, and depth-of-field requirements shift depending on where the camera sits on the car, which makes standardizing hard.
Newer approaches bring their own growing pains. Event-based sensors, for instance, don’t produce the kind of data conventional computer vision pipelines expect, so a lot of the real work now is in algorithms, not silicon.
And then there’s geopolitics and tariffs, US-China and EU-China trade tension, and OEM pressure to source outside China are pushing manufacturers toward places like Vietnam and Mexico. None of that is a technical problem, but it adds real cost and time to rollouts that would otherwise be purely engineering decisions.
Don’t miss Anas’ presentation ‘Automotive Image Sensors Market Dynamics, Trends and Challenges‘ at AutoSens Europe!
Interested in In-Cabin monitoring technology?
With a pass to AutoSens Europe, you’ll also get full access to our co-located sister event, InCabin. Take a look at the full agenda for InCabin here >>
