|
|
|
PC - LineScan
Back to Digital Frame Grabbers
Back to Coreco Frame Grabbers
| PC-LineScan is a high performance PCI-bus frame grabber that provides onboard
acceleration for high-speed vision tasks utilizing digital cameras. Designed
for challenging line scan applications, PC-LineScan includes performance enhancements
that collect and analyze large amounts of data in real-time and compensates
for variations in lighting. PC-LineScan integrates the real-time processing
functions needed for high-speed line inspection applications directly on the
board itself. This off-loads the host CPU and provides users with increased
speed and bandwidth to handle other tasks, such as motion control. |
 |
PC-LineScan is based on an expandable architecture that supports real-time processing
functions that works to accelerate machine vision tasks. This new design has been
fine-tuned for line scan applications. Features that were traditionally handled
more slowly by software are now available onboard in real-time, including Run-Length
Encoding and Flat-Field Correction. PC-LineScan also supports a plug-in expansion
module that can be custom designed to meet specific application needs.
Accompanying the new features of the PC-LineScan are several performance enhancements.
The 32-bit differential camera data interface now operates at frequencies up to
62.5MHz. This provides for continuous image acquisition at an incredible 250MB/sec.
Another enhancement to PC-LineScan is a larger 32MB SDRAM-based image buffer.
Finally, the PC-LineScan supports a plug-in daughter module to support additional
real-time processing functions.
Key features
- 120MB/sec bus master transfer to host that guarantees no missing lines and
a low host CPU burden
- On-the-fly re-sequencing of image data for multi-tap cameras
- Run Length Encoding gathers/compresses common data minimizing transfer time
- Flat-Field Correction compensates for lighting nonuniformities
- OPTO- 22 compatible parallel I/O (8/8) port eliminates need for an additional
I/O card
- Dual 16 x 16 LUT for image transformations
- 32 MB SDRAM image buffer supports acquisition rates up 250MB/sec (32 bits
@ 62.5 MHz)
- Scatter gather feature for high memory-efficient image transfers and management
Video Formats
- 32-bit differential LVDS input up to 62.5 MHz or RS-422 input up to 25 MHz
pixel acquisition rates
- Supports single-tap, dual-tap, or quad-tap monochrome or color digital cameras
Camera Data Inputs
- 32-bit LVDS (EIA-644) or RS-422 camera data interface at up to 62.5MHz (25MHz
for RS-422). Compatible with 8-bit (single, dual, quad tap), 16-bit (single,
dual) and 32-bit (single) digital line-scan and area-scan cameras. Also supports
10/12/14-bit image data (single or dual tap) and 8-bit color cameras
- Multiple camera support: up to three 8-bit cameras, up to two 10-bit, 12-bit,
14-bit, or 16-bit cameras, and one 24-bit or 32-bit camera. There is one set
of timing inputs; cameras must be externally synchronized
- Multiple boards can be synchronized for a wider multi-tap data width or more
multi-camera support
Camera Timing Inputs
- RS-422 or EIA-644 LVDS data and timing inputs
Differential timing inputs
- HDrive or Line Enable (LEN),VDrive or Frame Enable (FEN), and a Multipurpose
input (MULTI) for field, etc.
Programmable signal polarities
- Pixel Clock Input (PCLK): differential, programmable polarity edge sampling,
RS-422 0-25 MHz, and LVDS 0-62.5 MHz
- Line-scan and area-scan modes
External Trigger Inputs
- Two independent programmable triggers provide additional flexibility: start
and stop, sync and trigger, divide and trigger
- Programmable polarity - rising or falling edge triggered
- Software selectable sources for each trigger: software trigger, TTL input,
opto-coupled input, or differential input (same signaling as data/timing, LVDS,
or RS-422)
- Glitch detector/filter to condition noisy trigger sources
- The trigger rate-divider circuit supports acquisition at a user-selectable
division of the incoming trigger rate
Camera Timing Outputs
- The RS-422 or LVDS (EIA-644) are the same as data and timing inputs
- Output signals: Pixel Clock (CCLK), Camera External Synchronization (EXSYNC),
and Programmable Integration or Exposure Control (PRI)
- Deterministic multi-camera timing. EXSYNC and PRI can be synchronized to rising
or falling edge of CCLK
- Up to three identical sets of timing outputs controls up to three cameras.
Two dedicated sets of outputs. A third set may be enabled on the camera control
outputs
- Clock Output (CCLK): differential programmable frequency 320 KHz to 100 MHz.
Maximum frequency may be limited by signaling standard, cable quality, and length
- External Sync Output (EXSYNC): differential programmable polarity, programmable
period 210 ns to 18.75 s. User-programmable characteristics (i.e., pulse width,
pulse delay, etc.)
- Programmable Integration Output (PRI): differential programmable polarity.
User-programmable characteristics (i.e., start-point, stop-point, pulse width,
pulse delay, etc).
Camera Control Outputs
- Four logic outputs to control camera modes: the RS-422 or LVDS (EIA-644) are
the same as data and timing inputs
- Three outputs can be switched (by software) to timing output to support a
third camera
Flat-Field Correction
- Compensates for lighting nonuniformities (shading) and imager nonuniformities
via real-time pixel-by-pixel gain/offset corrections
- Corrects line-scan images up to 64K pixels/line
- Works with 8/10/12-bit image data
- Linear correction algorithm: y(DN) = x*{n(DN)-a}. DN = digital number; n(DN)
= raw data; a = dark offset (imager fixed pattern noise); x = correction factor;
y(DN) = corrected data
- 8-Bit data: n(DN) = 0-255 (8 bits); a = 0-63 (6 bits, 25% of full scale value);
x = 0-3.996 (2 integer bits, 8 fractional bits); y(DN) = 0-255 (results are rounded
to 8-bits)
- 10-Bit data: n(DN) = 0-1023 (10 bits); a = 0-1023 (10 bits, 100% of full scale
value); x = 0-3.9998 (2 integer bits, 12 fractional bits); y(DN) = 0-1023 (results
are rounded to 10 bits)
- 12-Bit data: n(DN) = 0-4095 (12 bits); a = 0-1023 (10 bits, 25% of full scale
value); x = 0-3.9998 (2 integer bits, 12 fractional bits); y(DN) = 0-4095 (results
are rounded to 12-bits)
- Operates at maximum data rates
- Processes up to 4 channels simultaneously for multi-tap camera applications
Input Look-Up Tables
- Two 16 by 16 Look-up Tables (LUT) can be programmed and used as four 8 by
8
- Perform point transformation before acquisition into image memory: threshold,
invert, absolute value, contrast adjustment, binarization
- Perform simple math based on 8-bit data input and constant, or based on two
8-bit data inputs: add, subtract, average, and multiply
- Normalizes 10, 12, 14-bit data to 16-bits
Run-Length Encoding
- Real-time "streak" feature extraction for line-scan applications
- Supports simultaneous RLE processing AND storage of the original image into
the image buffer
- RLE results is 0.5MB and may be stored in either the Feature RAMs or the image
buffer
- High-speed bus master transfers of data results from Feature RAMs to the host
- Feature RAMs support simultaneous RLE processing and access to data results
by the host
- 7-bit "Tags" allow up to 128 different features to be monitored
Flexible FPGA implementation may be downloaded with alternate configuration files
to support other processing functions such as:
- Gray Scale Thresholding
- Edge Detection and Gauging
- Other line-based processing algorithms
- Significantly reduces image data volume
- Processes both single-tap and multi-tap image data
- Operates at maximum data rates
Expansion Slot
- Supports a plug-in daughter module for real-time processing functions
- Functionally located between the pixel LUTs and the image buffer
- Full 32-bit image data pipeline
- Can function in parallel with the built-in RLE circuit
- Daughter card results may be transferred directly to the host or stored in
the image buffer
- High-speed bus master transfer of data results to the host is supported
Multi-Tap Image Resequencing
- Resequences multi-tap camera data in real-time: 2 tap, 4 tap, odd/even pixel,
odd/even line, opposite scan, multiple quadrant, etc.
- No host CPU intervention required (0 overhead)
Image Buffer Memory
- 32MB linear mapped SDRAM
- Acquires at camera data rate (up to 100MB/sec for RS-422 or 250MB/sec for
LVDS), transfers to host at PCI-bus data rate (up to 132MB/sec theoretical maximum)
- Color image data can be stored as 32-bit zero-filled or 24-bit packed format
- Programmable decimation by 2, 4, 8, 16; independent horizontal and vertical;
performed during acquisition, affects acquired image data
- Programmable "valid video window" timing - 1 to 64K pixels per line,
1 to 64K lines per frame
Output Scatter Gather Table
- "DMA Table" performs "destination scatter gather" for
bus master transfer from image memory to PCI-bus host
Bus Master Image Transfers
- Transfer rates up to 132MB/sec theoretical limit of PCI-bus, limited by bus
traffic, block size, and capability of destination device
- Bus master transfer to PCI-bus host memory, VGA, or other PCI-bus system resources
- Hardware "scatter gather" table provides efficient automated transfer
- Optional clipping during bus master transfer to VGA memory removes "color
dots" in Windows (Windows reserved colors)
- Optional padding 8-bit to 16-bit data during bus master transfer to VGA secondary
surface, for 4:2:2 display; also 16-bit to 16-bit padding shifts, 16-bit monochrome
to 8-bit adds neutral CrCb value
- Supports basic image transforms: flip, mirror, rotate by 180
- Supports "planar" 8-bit transfer of 24-bit color data during bus
master transfer
Parallel I/O Port
- One 16-bit parallel I/O port: 8-bit TTL input, 8-bit TTL/CMOS output
- Strobe input for latching input data, programmable rising or falling edge
- Strobe output for latching output data
- 50-pin header is pin-compatible with OPTO-22 module racks
Serial I/O Ports
- Two RS-232D serial ports on two 10-pin headers
- Software and register compatible with 16450 and 16550 devices
- l TXD, RXD, CTS, RTS on each port
- Data rates from 50 to 128K baud
Environmental
- Board Size - 312.0 by 106.8 mm; 12.283 by 4.205 inches (Full Card)
- Operating Temperature - 10-60 degrees Celsius
- Relative Humidity - 0-90% non-condensing
- Power Requirements (estimated) - 2.5 Amperes at +5 volts, 5 volt to 3.3 volt
conversion performed on the PC-Linescan
Sensor Interface
- Input - 2 RGB color or up to 2 sets of 3 synchronous RS-170 CCIR monochrome
cameras
- Programmable time base generator for standard and nonstandard video formats.
Resolution to 4K x 4K
- PLL or XTAL mode (Outputs horizontal/composite, vertical and reset timing
to camera. 3 sync sets available for driving 3 simultaneous mono cameras)
- Progressive scan (noninterlaced) signal supported
- Variable scan cameras are supported, single ended TTL timing (0 - 30 MHz)
- Dual-tap cameras are supported using two of the three RGB channels
Video Digitizer
- Monotonic 24-bit flash ADC; Input sample rate programmable to 30 MHz, 12Mhz
video bandwidth, 48DB SNR
Video Signal Conditioning
- Programmable input gain selection (4 gain factors)
- Programmable positive and negative signal reference: 0 to +5 Volts in 256
steps
- Video low pass anti-aliasing filter
- DC restoration: programmable clamp pulse
On Board Digital I/O
- OPTO-22 compatible
Color Space Converter and Look-Up-Table
- 10-bit precision 3x3 30 MHz matrix multiplier and a 64KDWORD (32 in - 32 out)
ILUT following ADC
Transfer Rates
- Less than 4 ms per image (Requires a motherboard with burst mode to host memory)
- Contact Coreco Imaging for suggested motherboards
External Trigger, Strobe Control, and Frame Reset
- Trigger provides synchronization to external events: 3 TTL with Schmitt Trigger
and (2) Opto-coupled, (1) RS-422 differential
- Flexible strobe control
- Support for frame reset cameras
- Edonpisha supported
Video Window Generator
- Programmable resolutions up to 4k x 4k. Area of Interest (AOI)
Interrupts
- Software interrupts can be generated at the start of the strobe pulse, the
start of a vertical blank, or at the end of a trigger cycle
Hardware Color Dot Clipping
Data formatting
- 8-bit Red, Green, or Blue
- 16-bit (converts 24-bit RGB to 16-bit YcrCb 4:2:2)
- 24-bit (packs 24-bit RGB to data into 32-bit DWORDS). This saves memory and
supports standard color VGA cards
- 32-bit standard RGBa or aRGB format
- 32-bit planar mode separates data from each 8-bit input into its own plane
in memory.
Display - Windows
- Display resolution as per installed VGA device driver
- A DirectX compatible SVGA adapter required for real-time display
Automatic Double Buffered Full Frame Acquisition
Bus Requirements
- 32-bit PCI slot
- 1.3 Amps @ +5 volts
- 0.2 Amps @ + 12 volts
- 8" long PCI card
Power Output
- 500 ma @ +12 volts per camera (1 Amp total)
|