Defining Measurement Error for Visual Inspection Applications
The methodology presented here is to determine the measure- ment error when the objective of an inspection task is to identi- fy freeform areas. This methodology was included in the work by the author.12
termination process, if the center position of another element from any of the element maps of the same part falls within this zone, then the two elements are assigned to the same master cluster region. Since the goal is to find overlapping and adjacent
The generic description of the methodology is
presented first, followed by an application to the metalcasting surface inspection task. The first step is for regular inspectors to identify the areas that are anomalous or otherwise of inter- est. Multiple, equally-sized circles, which are smaller than the area of interest, are used to approximate the size and location of the identified area. A series of tightly packed circles are placed until the area of interest is covered. When completed, a boundary line drawn around the circles would encapsulate the defective area identified by the inspector. In this paper, these circles are defined as elements. Figure 3 shows an example of how the method is applied to a part marked by an operator; the defective area is quantified by circles, or elements. For each inspection trial, an element map that contains the location of each element is constructed; each element map has the same coordinate system with respect to the part. Figures 4 and 5 show examples of castings at two different foundries which were inspected twice by two inspectors each, and then the in- spection marks were identified with circular labels.
Defining Master Clusters
To assess repeatability and repro- ducibility error, each part must be assessed at least twice by at least two operators. Each of these separate in- spections is defined here as a trial. This leads to an interesting situation, since the defect regions will likely be defined differently for each trial. For instance, a single defect region on one trial may be defined as two sepa- rate defect regions for a subsequent trial on the same part. This leads to the need to define a master cluster. A master cluster is a composite of all of the elements that would have con- tact with each other if the element maps from all trials for that part were superimposed. To determine the elements that comprise each mas- ter cluster, a search is conducted of all element maps for elements that would be adjacent or coincident.
The search for adjacent elements is done across all the element maps (one for each trial) for the part. The master cluster search zone is circular and centered on a particular element’s center. During the master cluster de-
Figure 3. A photograph of a casting with white stickers applied to identify the locations of the inspector’s marks. These stickers then become the elements used in the methodology defined here.
Inspection 1 Inspection 2
Figure 4. Results obtained from the same casting inspected twice by each of two operators at Foundry 2.
International Journal of Metalcasting/Summer 2011 9
Operator 2
Operator 1
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75