A prototype system for the detection, evaluation and mapping of defects on flat, uniform coloured fabrics has been carried out in the image processing field using a charge coupled device (CCD) linear camera and different configurations of light sources usually available on sale. After starting the research and mainly due to the difficulties in the collection of a sufficient number of defective samples on flat fabrics, it was agreed upon to start the research limited to the defects on the denim fabric. In the 5 years of research, 2 theoretical approaches to the detection have been identified and tested.
The first approach, derived from past experience in other industrial fields, was based on a passive detector. A common algorithm architecture was developed to detect the majority of flaws, even the small ones; to avoid complexity and to satisfy important constraints like real time and selling price. A statistical analysis of the detection has been made: about 30% of flaws could be detected and mapped without detection difficulties; about 25% of flaws are detected but with difficulties; about 45% of flaws are not detected at all. Conclusion: The results are not good enough for a marketable unit.
The second approach, using an active detector, is based on the fact that flaws are known. Indeed, human inspection is performed following a catalogue of known flaws; the working hypothesis in thus true. A new statistical analysis of the detection was made: about 95% of flaws were detected; more than 95% is expected if a digital image enhancement is carried out before the channel of detection. Some enhancers have been successfully tested in the laboratory. Due to the parallelism concept, speed inspection can be specified, ie increased with additional processors without changing the global architecture. Using a programmable kernel for the detection, applications on other textiles (like grey goods) can be easily implemented.
The automation of the process leads to higher inspection speed, reduction of cost through manpower saving and improvement of the inspection quality and consistency. The hardware flexibility and modularity of the detection system will permit the fast updating of the equipment when more powerful hardware and software solutions are available.
Although initial users are in the textile field, applications also exist in all fields where a continuous material has to be inspected in real time conditions (eg paper, plastic laminates, steel, etc).
In order to exploit these results, it is intended to design, construct and industrially test a prototype system for the detection, evaluation and mapping of denim fabric faults.
THE AVAILABILITY OF A COMPARATIVELY LOW PRICE MODULAR SYSTEM FOR AUTOMATIC ON-LINE DETECTION AND EVALUATION OF DEFECTS AND SHADE VARIATION MONITORING ON FINISHED FABRICS, ALSO CAPABLE OF SUPPLYING INFORMATION FOR THE COMMERCIAL EVALUATION OF THE FINISHED PRODUCT BESIDES AUTOMATION OF THE CUTTING AND SHIPPING PHASE AND NC CUTTING MACHINES IN THE GARMENT MAKING STAGE HAS BEEN IDENTIFIED AS A KEY FACTOR IN THE AUTOMATION OF TEXTILE MANUFACTURING INDUSTRIES.
HUMAN INSPECTION IS IN FACT WELL KNOWN TO BE COSTLY, UNRELIABLE AND INCAPABLE OF GIVING A STANDARD EVALUATION OF THE SAME TYPE OF FLAW SINCE THE ACTUAL JUDGEMENT OF A QUALITY CONTROL INSPECTOR IS ALWAYS STRONGLY INFLUENCED BY SUCH FACTORS AS, FOR INSTANCE, FATIGUE, OWN INTERPRETATION OF THE DEFECT STANDARDS AND SO ON. MOREOVER QUALITY INSPECTION IS NOWADAYS PERFORMED AT LEAST TWICE: DURING THE PROCESS OF TEXTILE PRODUCTION AT THE MANUFACTURER'S PLANT (OUTGOING INSPECTION) AND BEFORE CUTTING AT THE GARMENT MAKERS' (INCOMING INSPECTION).
THE MOST SIGNIFICANT ADVANCES ON CURRENT PRACTICES ACHIEVED BY INTRODUCING AUTOMATIC INSPECTION ARE: REDUCTION OF INDUSTRIAL COSTS THROUGH THE ELIMINATION OF MANPOWER FOR VISUAL INSPECTION; CONTINUITY, RELIABILITY AND REPEATABILITY OF JUDGEMENT IN THE EVALUATION OF FLAWS, WITH ADVANTAGES BOTH FOR THE MANUFACTURERS AND THE GARMENT MAKERS (FEWER CLAIMS FOR NON DETECTED DEFECTS); AVAILABILITY OF FLAW-MAPS DIRECTLY USABLE FOR AUTOMATIC CUTTING IN CLOTHING INDUSTRIES AND STATISTICAL QUALITY CONTROL AT THE MANUFACTURER'S; FINALLY THE POSSIBILITY OF USING INDUSTRIES AND STATISTICAL QUALITY CONTROL AT THE MANUFACTURER'S; FINALLY THE POSSIBILITY OF USING THE SAME EQUIPMENT FOR GREY GOODS AND INCOMING INSPECTION.
Funding SchemeCSC - Cost-sharing contracts