Monday, February 14, 2011

Melexis Introduces Automotive HDR Sensor

PR Newswire: Melexis launches the MLX75411 "Avocet" HDR sensor designed for driver assistance and night vision applications. The sensor will also be beneficial for security/surveillance cameras, fleet safety cameras, truck blind spot cameras, transportation/heavy vehicle cameras, and other outdoor and all weather cameras.

The Avocet image sensor is 1/3-inch format wide-VGA sensor with a 1024x512 pixels resolution, using 5.6um pixels. It features an extended high dynamic range of up to 154dB in every single frame at a full frame speed of 60fps. The sensor is based on SmalCamera-Cypress-Sensata Autobrite technology. It also has on-chip histogram optimization and sharpness control.

Cliff De Locht, product line manager at Melexis says: "In specific daytime automotive conditions like when sunlight shines directly into the lens, or when driving in or out tunnels or garages, the "Avocet" image sensor is able to give a crisp picture with maximum picture details in the very bright areas of the frame and the dark areas of the frame simultaneously."

The MLX75411 "Avocet" samples are available in 4 versions: monochrome, color RGBG, color RGBi and RCCC.

14 comments:

  1. What is the meaning of 154dB? What kind of situation on road can result in such dynamic range? I imagine that from 0.1lux (moon light) to 100 000Lux (sun shine), 120dB is enough.

    ReplyDelete
  2. What about "no moon light" -> >120

    ReplyDelete
  3. i think they are trying to pick up IR as well, so maybe than be < moonlight (0.1 lux)?

    ReplyDelete
  4. 1024x512 sounds completely not VGA or even similar. This is a wide format 2:1 sensor.

    ReplyDelete
  5. Have you seen 3T CMOS sensor which can see below moonlight?

    ReplyDelete
  6. Can they display the 154 dB image, given that the dynamic range of most displays is much less? Secondly, by dynamic range do they mean the sensor dynamic range or the system dynamic range (inlcuding the lens system which limits the DR to ~100 dB)

    ReplyDelete
  7. Let's calculate:
    The frame time = 16.666ms, suppose that this is allocated totally to the dark areas in an image, then 154dB (10^7.7), the equivalent exposure time on the brighest non-saturated areas should be 0.3ns. Is this possible? Suppose that they can put a gain of 10x in the readout chain, they need still 3ns exposure time for the brightest area.

    Conclusion?

    ReplyDelete
  8. To the last anon: You make mistakes with your calculations...you must consider the DR enhancement instead of the total DR when you calculate the min integration time. If the original DR of the pixel without enhancment techniques is 60dB then DR enhancement=94dB. It follows that min integration time is ~300ns. With a gain of 10x you get 3us as min int time. That is reasonable!!!

    ReplyDelete
  9. @Adi, DR=max_sig/dark_noise, so the gain can not play any role here in terms of DR. So you have to shorten the exposure time to add 94dB more DR. So effectively the minimum exposure time is 300ns, but not 3us.

    ReplyDelete
  10. You are right...my 3us was compared to the 3ns at 10x gain of the previous comment.

    ReplyDelete
  11. To answer the question about displaying 154dB, you have to think about tone mapping the 154dB to 48dB for display. Anyway this kind of sensor is also used a lot in image processing applications, not in display applications.

    ReplyDelete
  12. Why 154 dB? - because you need to differentiate objects under moonlight conditions at the same time you need to differentiate brightly lit objects. For example - you need to differentiate the white lane marker from the gray pavement under moonlight conditions at the same time you need to differentiate the difference between a headlight and tail light.

    ReplyDelete
  13. it's impossible to have such optical DR at the focal plan. How have Melexis guys measured this?

    ReplyDelete
  14. Can this sensor see any under moonlight conditions at 60fps???

    ReplyDelete

All comments are moderated to avoid spam.