LINESCAN IMAGING: High-res cameras use fiber interfaces

Emerging from the Austrian Research Centers (Vienna, Austria), Smart Systems (Vienna, Austria; www.smart-systems.at) develops a range of technologies and products targeted at machine vision and image processing. The company, which announced a series of products at VISION 2008 in Stuttgart, Germany, surprised attendees and exhibitors alike with a range of products that included the world’s fastest linescan camera, the world’s first 10-GigE camera, and a specialized smart camera that included the company’s own in-house developed CMOS imager.

“To achieve linescan rates of 82 kHz,” says Ahmed Belbachir, a research scientist with Smart Systems, “the xposure 4000m linescan camera does not use a conventional linescan imager in its design. Rather, the camera incorporates a 2352 × 1728-pixel array sensor from Aptina Imaging (San Jose, CA, USA; www.aptina.com) and reads just a few lines (4, 6, …, 16) from the imager to attain these speeds. In the color version of the camera, the xposure 4000c, a few Bayer-filtered double-lines are read from the imager to produce a color linescan image.

“Because the camera uses an area-array sensor in its design, it can be configured as a linescan camera or, alternatively, a time-delay integration (TDI) mode-like camera, or an area-array camera,” says Belbachir. “In a multiple exposure (MX) method, several time-delayed images of the same object can be synchronized to object movement, exposed, and read out, corresponding to the number of MX stages or lines used. This effectively extends the exposure time of the camera.”

The xposure 4000 cameras are the second generation of the xposure family. At VISION 2007, the xposure 1300 cameras were introduced. Also based on Aptina CMOS imagers, this camera series has a resolution of 1280 × 1024 pixels. Like their high-resolution counterparts, the xposure 1300m and 1300c cameras can also be operated in both linescan and area-scan modes. When used in linescan mode, these cameras can output linescan images at 100 kHz, the fastest currently available from any linescan camera vendor. To speed image-processing functions, flat-field conversion, color space correction, and lens distortion correction are all performed using an on-board FPGA.

Because of the 10-m cable limitation of the current Camera Link standard, Smart Systems has also developed one of the world’s first 10-GigE camera interfaces for its camera series. At VISION 2008, the company showed the first version of the interface embedded into its xposure 10G camera (see figure). In essence, this camera uses the design in the xposure 4000 camera and couples the Camera Link Medium output to a high-speed 10-GigE fiberoptic interface.

Smart Systems has demonstrated the first 10-GigE camera, the xposure 10G, which couples the Camera Link Medium output of its 2352 × 1728, 82-kHz series of cameras to a high-speed, 10-GigE fiberoptic interface.

To configure these cameras, the company offers xShell, a TCL-shell derivative and a camera library (xLib) that allows the developer to configure exposure times and image-processing functions such as flat-field correction. According to Belbachir, a GenICam interface can be built onto the xLib but is currently not under development. At present, Smart Systems is looking for an exclusive partnership concerning production and sale of the xposure 4000 camera series.

To add to the company’s portfolio of products, Smart Systems has also developed a smart camera based on its own industrial vision sensor. This 256 × 2 linescan imager is unusual in that it incorporates logarithmic photoreceptors (to attain a 120-dB dynamic range), and pixel-level analog data preprocessing. On-chip high-precision timing circuits, address generator, and bus arbiters are then used to format the data, which are buffered in an off-chip FIFO before being processed by a Blackfin DSP from Analog Devices (Norwood, MA, USA; www.analog.com).

“Because the Industrial Vision Camera sensor uses on-chip asynchronous analog preprocessing of the visual information,” says Belbachir, “output data are only generated in the pixels if relevant information is present, thus highly reducing redundancy and data rate. This is especially useful in machine-vision applications where very high temporal resolution is called for.” Indeed, the company’s smart camera has already been tested in such an application at SIG (Saalfelden; Austria; www.sig.biz), a manufacture of cartons for beverages and food products.

“In the production of beverage cartons,” says Belbachir, “it is desired that the correct amount of beverage packages squeezed into flat sleeves is packed in each box. If too few are packed, the customer may return the carton; if too many are packed, then this will result in added expense for the carton provider.” In one application evaluation at SIG, boxes of nominally 350 cartons are moved along a production line at a rate of 0.5 m/s. At these speeds it is obviously impossible for an operator to manually count these cartons. Using Smart Systems’ smart camera, the cartons can be reliably detected under difficult and uncontrolled lighting conditions and the carton count returned to the automation system. Although this smart camera is not actually deployed at SIG, this test system demonstrated its usability for high-speed counting.