|Most laser scanners sweep the laser beam horizontally using an electronically controlled oscillating mirror. |
Laser scanners use a moving pinpoint of light to illuminate the barcode. A single photocell receives the reflected light and converts the barcode into an electrical signal as it moves across a barcode. The scanner then measures the relative widths of the bars and spaces, translates the different patterns into the barcode data, and sends them on to a computer or portable terminal.
Laser scanners are able to read one-dimensional (1-D) barcodes in traditional high contrast applications, providing excellent performance for high-speed unidirectional scanning, offering large depth-of-field and a simple setup. Stitching several lasers together in an "X" or star pattern allows for omnidirectional scanning. However, less than ideal conditions — poor contrast codes, poor code quality, and perspective (angled) reading — as well as the need to read the increasingly popular Data Matrix and other more complex two-dimensional (2-D) matrix codes — require a more robust approach. Although laser-based systems are easy to set up, it can be a challenge to maximize read rates because users cannot see the images that the scanner is attempting to interpret. In addition, it is often difficult to determine the reason for an unsuccessful read, so optimizing the process proves next to impossible.
|In conventional image-based reader technology, the horizontal width and a vertical height which define the scene the camera can see. The camera field of view gets wider with greater distance from the lens but each pixel covers more area. |
The other choice: conventional image-based readers generally connect commercially available components such as CCD or CMOS imagers and digital signal processors (DSPs) on a conventional printed circuit board (PCB). The reader captures an image, passes that image to the DSP to locate and decode any barcodes present, and then captures the next image in a pipeline fashion. A production environment with well-formed high contrast 1-D barcodes can achieve reading speeds up to 45 decodes per second — suitable for many high speed applications that can benefit from the more robust reading capabilities of image-based systems.
Advantages of Imagers
Image-based readers use solid-state camera technology to capture full-frame images in a single snapshot, avoiding the need to cross a barcode with scan lines from end-to-end the way lasers do. When an imager captures a frame containing a barcode, it can locate that code anywhere within the frame. Thus, in effect, each image contains thousands of scan lines in all directions making finding and reading barcodes more robust. Image analysis software algorithms can interpret a higher percentage of the captured codes than laser scanners can, and they can also manage the complexities of 2-D codes. In addition, the images themselves can be stored for training, historical analysis, or documentation.
Conventional image-based readers are typically designed for a single focal plane and offer only limited depth of field. System setup requires fixing variables such as the focus of the lens, aperture size, and target brightness. Changes in distance after setup cause images to be out of focus, overexposed, or otherwise misaligned. When a process provides consistent focus and lighting, however, conventional image-based readers offer clear benefits. ID readers that incorporate liquid lens technology with autofocus, allow users to change focal distances easily during line changeovers, thus reducing laser scanners' depth of field advantage. Also, software tools make optimizing a setup easier with features such as automatic target brightness and robust image analysis algorithms that are designed to handle some contrast variations and lighting changes.
Applications requiring speeds greater than 45 decodes per second require increasing image capture rates, processing speeds, and throughput. Decode speed — defined as the time required to capture an image and analyze it — is currently limited by the distance between the imager and the processor, the associated data transfer rates, and primarily the heavy analysis burden on the single DSP. Combining the imager and processor on a single piece of silicon permits handling two tasks almost simultaneously, dramatically increasing decode speeds and overall frame rates. With its proprietary VSoC on-chip vision system, the company has achieved this level of integration with the DataMan® 500 barcode reader.
|DataMan 500 image-based reader. |
By capturing images and processing them in parallel, the on-chip vision system enables image capture speeds up to 1,000 frames per second and effectively doubles image-based 1-D barcode reading speeds to 90 decodes per second. A typical system with separate imager and processor operates in a serial pipeline fashion (one trigger, one image, one finder, one decoder, one output).
Image Capture Speed
For example, if a typical system runs at 60 frames per second, then an image capture occurs every 16ms. That image proceeds to the processor, which must find and decode it in just 16ms before it receives the next image. If the processor takes longer than 16ms, then images must queue in a buffer and wait to be processed, reducing system throughput.
In contrast, the on-chip vision system captures multiple images through an autoexposure routine and determines in which image a barcode is present, locates it in the field of view, and reports its coordinates and orientation "on the fly" to the DSP.
When the on-chip vision system sends the images to the external DSP, it identifies which images to process and which to ignore. The external processor does not waste valuable time analyzing images that do not contain barcodes. For those that do, knowing the coordinates of the barcode means that the processor need only decode those small portions of the image, dramatically reducing the DSP's processing load. The on-chip vision system effectively increases the image capture frame rates and the system throughput at the same time. "One trigger, one image, one finder, one decoder, one output", becomes "no trigger, multiple images with finding, one decoder, one output."
Simpler, Easier Setup
Increasing the frame rate without causing a huge buffer backup in the DSP offers other advantages as well. Higher frame rates combined with large pixels mean shorter exposure times. Shorter exposure times and large pixels mean the system can achieve higher speeds with less light than conventional imagers and can produce crisper, clearer images in high speed applications.
|Conventional serial pipeline. |
The typical "one trigger, one image, one finder, one decoder, one output" scheme is particularly sensitive to variations in line speed or part location. The trigger must be well synchronized with the camera so that the barcode is in the same position in the image every time. On the other hand, the on-chip vision system's high frame rate and its ability to discard images without codes as well as pinpoint the codes in the remaining images eliminate this need for a trigger. Some conventional high-speed imaging applications use an advanced triggering technique called "burst mode" to overcome positional uncertainty. For example, some image-based readers can capture up to 22 images at a frame rate of up to 60 frames per second. Of the 22 images, perhaps only two or three that contain the barcode are buffered in the reader. Decode timeouts, set appropriately to decode those two or three images, would apply to all 22 images. Parts moving toward the reader would have to be spaced so that the reader finishes processing the buffered images before acquiring the next set of 22. Although this approach addresses positional uncertainty, the necessary part separation introduces an additional process complication.
By contrast, the on-chip vision system's much higher frame rate allows capturing images without buffering them or requiring part separation. Also, the higher frame rate captures many more images containing the barcode, permitting more opportunities to read each code. In other words, less sensitivity to position uncertainty offers a more robust read result.
|On-chip vision system decodes at double the speed compared to the pipeline system. |
Even with a damaged code or in less-than-ideal lighting conditions, at least one of the captured images will likely permit a relatively easy analysis. In addition, processing timeouts for images with codes can be set toward the average rather than the maximum, again improving total system throughput.
For straightforward applications that require only reading well-printed 1-D barcodes at high speeds at fixed positions, laser scanners will likely continue to provide a reliable, cost-effective solution. But when the more advanced capabilities of image-based scanning are needed, along with high throughput, on-chip vision systems offer the best — if not the only — viable solution. This next generation technology reduces downtime and changeover time, and will result in fewer parts rejected, less manual rework, and cost savings.
Contact: Cognex, One Vision Drive, Natick, MA 01760-2059: 877-264-6391 fax: 508-650-3344 E-mail: firstname.lastname@example.org Web: http://www.cognex.com