 |
|||
| Dyeing and finishing, being an end of the line process for a textile, traditionally bears a large proportion of the cost of faults. In order to minimise customer rejections and returns dyers and finishers are forced to suffer the high labour costs associated with detailed manual inspection. And manual inspection is not always a guaranteed solution to the problem. |
|
||
| The small size of the defects, which need to be detected, can only be seen by a human if the textile is travelling at a slow speed. This slow throughput of an individual inspector relative to production speed requires several inspectors working in parallel. With each batch being inspected by different people the inspection standard will vary from batch to batch. And with the inspectors attention waxing and waning throughout the shift inconsistencies will be found in individual batches. |
|||
The inconsistencies can be further aggravated by labour turnover that will add further variation to the outputs. |
|||
| Utopia would be an inspector that never sleeps, applies consistent inspection criteria and does so at the production speed of the process. There are a number of vision systems on the market, some being capable of inspecting textiles. Of those that are capable of inspecting textiles, the majority can inspect either texture defects or colour, but not both texture defects and colour. |
|||
|
Computer vision systems are large and complex, typically involving multiple planes of cameras with several cameras on each plane and numerous lighting arrangements. The operation of such machines has, in the past, been an unwieldy task. Traditionally, the process of setting up such systems has been long and arduous with multiple sets of parameters needing to be set for each differing material, colour, customer or batch number. |
||
The technologies available on the market today vary widely. Using the lower technically specified systems still requires a considerable amount of set up and manual interaction. At the higher end of technical sophistication the vision systems have self-learning and the ability to run and re-run webs or cloth in a simulated fashion thus reducing the need for manual interaction and set up times. |
|||
| One of the high specification systems is the Shelton family of vision system platforms. The Shelton webSPECTOR ® family includes Lite, Standard and Plus versions of the system concept. Each module tackles a different area of the automated vision inspection arena: - |
|||
| webSPECTOR ® Lite is designed for measurement and gauging operations where the features being looked for are a priori and exist to some extent in a controlled way. |
|||
| webSPECTOR ®Standard tackles defect finding tasks were the feature being looked for is not known but simply differs from what should be there. |
|||
| webSPECTOR ® Plus offers enhancements such as self-learning, data base management, defect libraries and the webcorder utility. |
|||
| The webSPECTOR ® family uses MMX technology in most of its processes. This means there is considerable power available to undertake the different image process requirements that identify the subtle defects that would otherwise be missed and sent to customers. The webSPECTOR ® Standard and webSPECTOR ® Lite systems use nine different defect detection algorithms that can run simultaneously. Six of these use types of filtering which are very “process-hungry” operations but give the accuracy that is required. The other three processes are image stabilisers and cancel out external effects such as variable lighting and factory temperature. This gives better reliability so that the system can inspect to the optimum sensitivity. |
|||
Page Select 2 |
|||
 |
|||
