eeTimes
eeTimes
eeTimes eeTimes EDN
Forgot password Register
Print - Send - -

Technology News

Analog Devices optimizes processors for embedded vision

March 28, 2012 | Christoph Hammerschmidt | 222902147
Analog Devices optimizes processors for embedded vision Advanced Driver Assistant Systems (ADAS), industrial machine vision applications and surveillance cameras have one thing in common: They have to process large amounts of image data in real-time. Analog Devices has developed a new series of its Blackfin processor family that has the capability to greatly speed up these processing steps.

The new dual-core processor types BF608 and BF609 are equipped with a powerful image processing accelerator, the Pipelined Vision Processor (PVP). This device enables the processor to run up to five concurrent vision algorithms. In an ADAS application, these could be, for instance, pedestrian detection, collision warning, lane detection, traffic sign recognition and automatic high beam / low beam switching. In addition, the device can estimate the distance of other objects - not by measuring it but by applying plausibility algorithms to moving objects. 

Fig. 1: The performance of the Pipelined Vision Processor enables ADI's new BF60x processors to perform multiple video analysis tasks concurrently

This high data processing capability along with the extended operating temperature range of up to 105°C required for automotive applications and hardware-based functional safety according to ISO 26262 makes the two new chips ideal for automotive deployment, explains Peter Voss, ADAS products marketing manager for Analog Devices GmbH. The market environment seems to be favorable: While currently ADAS are sold as optional equipment and in relatively moderate numbers, this is likely to change in the midterm. The reason is that the EuroNCAP car safety performance assessment program has significantly increased its requirements for pedestrian protection. Cars that do not meet these requirements cannot achieve the best rating anymore.

One of the most popular measures to improve active pedestrian protection are camera-based warning systems. However, to achieve the best EuroNCAP rating, these systems need to be part of the basic configuration of the vehicle. This will significantly push the market penetration of such systems, believes Voss.

In order to enable small packages and high packaging density, the new processors are optimized for low power dissipation. At room temperature, the devices get along with just 400mW; at maximum operation temperature of 105°C they still do not exceed a moderate value of 1.3W. Other features include a powerful switched-fabric based data transfer infrastructure inside the chips and a rich selection of interfaces. The two processor cores can access up to 4.3 MBit on-chip SRAM. 

Fig.2: The block diagram of the BF60x unveils a rather complex inner life.

Both the BF608 and the BF609 can process up to 30 frames per second. The difference between the two is that the BF609 processes images with HD resolution while the BF608 only is capable of processing VGA images.

The product rollout is complemented by two Blackfin models with similar configuration but without the high image processing capability. These models, the BF607 and BF606, do not dispose of the Pipelined Vision Processor. With these processors, ADI targets general-purpose DSP applications including wireless communications, industrial controls and grid monitoring, Voss explained.

For all new processors, ADI will provide a full development toolset including Eclipse-based CrossCore Embedded Studio.

Samples of the new Blackfin embedded vision processors are available. Volume production however will start only in March 2013.

For more information, visit www.analog.com/Blackfin60x

 










Please login to post your comment - click here
Related News
MOST POPULAR NEWS
Interview
Technical papers
Poll
Which technological approach contributes most to sustainable mobility?

All material on this site Copyright © 2009 - 2010 European Business Press SA. All rights reserved.
This site contains articles under license from EETimes Group , a division of United Business Media LLC.