Image analysis – Applications – Manufacturing or product inspection
Reexamination Certificate
1999-05-07
2003-07-22
Ahmed, Samir (Department: 2623)
Image analysis
Applications
Manufacturing or product inspection
C348S127000
Reexamination Certificate
active
06597804
ABSTRACT:
FIELD OF THE INVENTION
The invention relates to a method for testing the reliability of a testing apparatus which checks a large number of objects of the same type for one feature, generates a feature signal for each object and checks the feature signal for fulfilment of a first condition, a test object being conducted to the testing apparatus after a number of objects and the feature signal of the test object being checked for the fulfilment of a second condition.
BACKGROUND OF THE INVENTION
According to the state of the art, the procedure in reliability tests for testing apparatuses e.g. those for empty drinks bottles, so-called empty bottle inspectors, is that a row of test bottles is prepared so that each contains a certain defect according to the defect recognition specification, i.e. does not fulfil a certain feature. A special test bottle is prepared for every feature checked. The batch of test bottles is then incorporated in the stream of bottles at certain time intervals, e.g. every half hour, or after a certain number of bottles, e.g. 50,000 bottles. This operation can be automatic or manual. The test bottles are marked so that they are instantly recognisable as test bottles. The reliability test consists of checking whether these test bottles can be recognised as defective by the testing apparatus, for example the empty bottle inspector. In the method used up until now, the second condition is thus complementary to the first condition, i.e. the second condition is fulfilled in the test bottles if the test apparatus recognises that the test bottle is defective, i.e. that the first condition is not fulfilled. A record is kept in the testing apparatus. If the reliability test fails, i.e. if one or more of the test bottles is not recognised as defective, the test must be repeated. This is to ensure the operational dependability, i.e. the reliability, of a testing apparatus. This reliability test is unsatisfactory as it is only subsequently recognised that a testing apparatus has no longer functional reliably. The reasons for the failure of a testing apparatus are usually a dirty lens system or a failure of individual components of the recognition electronics.
With the method according to the state of the art, a large number of test bottles had to be used to test the reliability for example of an empty-bottle inspector, every test bottle having a single defect, e.g. a defective closure thread or a foreign body in a single recognition zone. Every type of defect and every recognition zone thus required its own specially prepared test bottle. If a test bottle had revealed several defects, the fact that this bottle was singled out would not have ensured that all the defects were recognized. A set of test bottles therefore consists of e.g. some 10 to 15 bottles.
SUMMARY OF THE INVENTION
The object of the invention is to create a method whereby a deterioration in the operating performance of a testing apparatus can be recognised as early as possible.
According to the invention, this object is achieved in that the second condition is that the feature signal of the test object corresponds to a previously entered reference value.
The idea underlying the method according to the invention that of testing the reliability of the testing apparatus by means of a test object whose feature signal when the test apparatus operates correctly is very precisely known and is entered in the testing apparatus as a reference value. Apart from unavoidable inaccuracies when checking the test object, the feature signal must correspond exactly to the reference value. It does not matter whether the feature signal fulfils the first condition, i.e. whether the test object is free from defects.
For non-defective objects, the feature signal can lie within a larger range or above or below a threshold value.
The method according to the invention is suitable in particular for testing the reliability of empty bottle inspectors.
A procedure frequently used with empty-bottle inspectors when checking on the absence of foreign bodies in that an image of the object is scanned pointwise e.g. by means of a CCD camera, generally in two directions at right angles to each other, and the brightness of each image point is ascertained and light-dark and dark-light transitions recorded by comparison with the brightness of adjoining image points. Such a transition always occurs e.g. if the scan passes over the edge of a foreign body in an empty bottle. Even empty bottles which are free from foreign bodies have a certain number of brightness transitions, e.g. up to 100 brightness transitions, due to uneven areas in the receptacle wall or the fluting on the edge of the base. A single object counts in this case as free from foreign bodies up to 100 brightness transitions, i.e. a feature signal of 100 still satisfies the first condition.
During proper operation e.g. of an empty bottle inspector, a feature signal is obtained for the predominant majority of the empty bottles e.g. 90%, which is somewhat below the number of 100 light-dark transitions. If the sensitivity of the recognition device of the testing apparatus drops due to dirt or other reasons, this tends to lead to a decrease in the number of light-dark transitions recognised per empty bottle. If the reference value of a test bottle is e.g. 95, the feature signal of the test bottle falls with a decline in the sensitivity of the recognition device. Depending on how large the deviation from the reference value is, various measures can be taken. For a deviation of 10% e.g. a warning signal can simply be given out, while for a deviation of 20% or more the testing apparatus and the entire transport apparatus can be stopped.
A particularly advantageous version of the method according to the invention results in conjunction with defect recognition methods in which not only the number of light-dark transitions are counted, but also the light-dark contrast of the brightness transitions is established. The found image elements deviating from the background are divided into e.g. eight different brightness classes or the light-dark transitions are divided into e.g. eight different contrast groups, the number of light-dark transitions being counted only after this classification and being compared within each class with a special threshold value. If the recognition device of the testing apparatus is dirty, a light scatter appears on the lenses or the glass protection disks, which leads to a decrease in the light-dark contrast, as a misty-like blurring effect covers the image scanned by the recognition device because of the light scatter. The reduction in the light-dark contrast causes a shift in the light-dark transition within the contrast classes, in such a way that the feature signals more frequently fall in the classes with less contrast. When testing the reliability of the testing apparatus, there is now no need to rely just on the comparison of the number of light-dark transitions, but a comparison is carried out with a threshold value in every one of the contrast classes and, in addition, the distribution of the feature signals over the individual contrast classes can be taken into account.
Assuming that a test bottle has 100 light-dark transitions of the order of 250 shades of grey, and two contrast classes are used, the first contrast class containing light-dark transitions with under 180 shades of grey and the second contrast class containing light-dark transitions with over 180 shades of grey: if the lens system of the recognition device is slightly dirty, then 100 light-dark transitions are still recorded, but with diminished contrast, e.g. only 150 shades of grey. The established feature signal then reads “100 light-dark transitions in the second contrast class” and thus does not correspond to the reference value which reads “100 light-dark transitions in contrast class 1”. The deterioration in the possibility of defect recognition by the recognition device thus becomes noticeable by shifting the light-dark transitions from contrast class 1 into contrast class 2. Throu
Ahmed Samir
Bali Vikkram
Gardner Carton & Douglas LLC
Heuft Systemtechnik GmbH
LandOfFree
Method for testing the reliability of a testing apparatus... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for testing the reliability of a testing apparatus..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for testing the reliability of a testing apparatus... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3082459