USB 3.0: Addressing New Challenges in Machine Vision
- Created on Saturday, 01 December 2012
Users of machine vision systems often have one common goal in mind: increasing system efficiency. Greater efficiency translates into high productivity. On the factory floor, higher speed in an automated optical inspection system, for example, contributes directly to profit.
Greater efficiency can directly be correlated to the higher speeds of imaging systems. While machine vision tasks certainly require high-quality, high-resolution images and more compact systems, the universal requirement in machine vision is faster imaging speed.
Machine vision was born on the factory floor, increasing product quality by delegating tedious and repetitive inspection tasks to computer-based systems. The benefits of machine vision have now found numerous applications beyond the factory. While the factory floor can be a demanding environment in its own way, machine vision in transportation also poses its own unique integration challenges. These systems face uncontrolled lighting conditions and have tight physical restraints, operating over a wide range of conditions.
Even on the factory floor, smaller and easier-to-use systems are in high demand. Factories must promptly adapt to a rapidly changing environment driven by global competitive pressures. Price pressures also favor simpler and easier-to-integrate systems with reductions in both installation and operation costs.
As the demand for faster systems increases, however, a new generation of imaging sensors delivers images of higher resolution and frame rates, unattainable just a few years ago. In particular, CMOS sensors with global shutter are proving very attractive in machine vision. The imaging sensors deliver frames rates close to 400 frames per second at megapixel resolution, combined with 10- and 12-bit depths. The advanced sensors, which produce a large amount of data, put even more pressure on the bandwidth capability of the data interface.
Benefits of USB 3.0 to Machine Vision
The USB 3.0 specification, which offers considerable performance improvements over USB 2.0, is a response to the demand for a high-speed, high-bandwidth computer peripheral bus. Computer users who manipulate large files, such as movies or enormous collection of high-resolution images, have been the first to benefit from USB 3.0’s ability to provide the high bandwidth needed for fast data transfers.
Integrators looking to build a highspeed, high-resolution machine vision system can also benefit from this technology. Several industrial cameras manufacturers acknowledged this trend and have or will be releasing product lines of USB 3.0-based cameras to meet this need.
As its name implies, USB 3.0 is a technology step above USB 2.0, one of the most widely used data interfaces. One of the most compelling features of the USB 3.0 interface for machine vision systems is its high bandwidth. USB 3.0 provides an effective transfer speed of approximately 5 Gbps, which is ten times faster than USB 2.0 and five times faster than the widely deployed Gigabit Ethernet (GigE) interface. USB 3.0’s data transfer speed is effectively approaching Camera Link and CoaXPress speeds. Unlike Camera Link or CoaXPress, however, USB 3.0 does not require any special interface cards or frame grabber in the host computer.
Another critical benefit of USB 3.0 over USB 2.0 is the increase in computing efficiency. Without compromising on higher transmission speed, the USB 3.0 protocol allows for more efficient, resource- friendly data transmission. By supporting the use of Direct Memory Access (DMA), USB 3.0 host controllers are able to retrieve image data from USB cameras with minimal CPU involvement, reducing computer loading and freeing up resources for mission-critical algorithm processing.
USB 3.0 delivers 4.5 watts of power to a device, about twice the power of USB 2.0. The increase is coupled with more efficient power management techniques, for example, the elimination of the power-wasting polling mechanism. USB 3.0 has enough power to drive most machine vision cameras right off of the USB port power.