Product Recognition and Traceability
Product recognition makes it possible to distinguish one product type from another. Product traceability produces a record of one unit's manufacturing history. Product recognition and traceability have become a key component for manufacturers to manage material flow and reduce risk of potential liability.
As consumers become wary of recalls and product marketing claims, Xiris’ OCR & OCV solutions ensure that products have the correct identification, and are delivered to the proper staging area. Not only does this help reduce processing cost, Xiris’ OCR & OCV solutions help manufacturers protect brand equity and build consumer trust.
PRT1000 OCR & OCV Solution
Xiris employs a decision-theoretic approach in which the key features of every character to be recognized are extracted and saved as prototypes during a training operation. Then in recognition mode, an unknown character is identified as being the one in the stored reference set that provides the closest match.
Main Features:
| Accuracy | PRT1000 can delivery near 100% read accuracy on good characters. Similar performance is achievable on mediocre characters with additional library options. During training, options can be employed to specify which characters are allowed to occupy each position in the character string. |
| Fonts | PRT1000 has pre trained files for OCR-A, OCR-B, semi OCR and 5x9 dot matrix font. The read characters can be uniformly or proportionally spaced, continuous stroke, dot matrix. The PRT1000 also supports training of user specific fonts. |
| Contrast | PRT1000 can accommodate low contrast conditions and varying contrast conditions. The PRT1000 employs non linear or adaptive thresholding to locate and isolate the characters from the background image. |
| Degradation | PRT1000 can be pre-trained to recognize degraded characters that are typical for a particular manufacturer. The PRT1000 supports characters that are touching. An additional library set allows for identification of character strings with intra-character spacing greater than the inter character spacing. This is occurs most often with output from inkjet, dot pended and pin stamped output. |
| Scale | PRT1000 reads characters appreciably larger or smaller than those used in training as well as ones that exhibit altered aspect ratios. |
| Orientation | PRT1000 can read characters and strings that are off axis. With pre-training, the PRT1000 can also read rotated objects. |
| Data Matrix Bar Codes | The PRT1000 can provide 1D and 2D reads for date codes, bar codes and data matrix codes Data matrix and bar code applications use similarly developed libraries and as such the solution will handle contrasts, orientation, scale, and identify the location. The library tools allow for reads and identification of the common symbologies. |
Examples:
|
Multiple ROI reads from an inkjet printer on a beverage container |
|
Read of symbols trained as patterns |
 
|
Data matrix read from an image with low contrast ratios |
|
Dot peen read from a metallic surface |
|
Read of an inkjet label with different scale |
|
Alphanumeric read of characters at different angles |
|
 |
|
 |