IMcProfileEdgesSizeWeighting Property

Property Value

Type: Double

When FindEdges is called the mcpefmMatchPattern EdgeFindingMethod performs a correlation between the ProfilesAncestor profile samples and the Pattern. This correlation is then adjusted for the amplitude of the matched profile segment to come up with the weighted profile value for that profile sample segment. The strength of this amplitude adjustment is set by the SizeWeighting property, which takes on values from 0 to 100. A value of zero gives no size weighting at all, so that only the degree to which the profile matches the pattern's shape matters. A value of 100 (the maximum allowed), strongly favors larger amplitude's where the profile matches the pattern. As a practical matter, zero is almost never a good choice, because even the smallest noise variations in the profile can give a good match. The default value of 50 gives a modest degree of size independence, without overemphasizing noise. When the WeightingFlags mcpewfPointsForSameSize bit is set, then amplitude analysis is done somewhat differently. In this case, the heighest weights are given to matching profile patterns where the amplitude of the profile variation most closely matches the amplitude variation of the Pattern property. The strength of this size matching is controlled by the SizeWeighting. A value of zero means that no size matching is performed (that is, only shape matching matters). A value of 100 gives very strong size matching, so that only profile segments that almost exactly match the Pattern amplitude will get much weight. The default value of 50, gives moderate weights to matching sizes, so that profile sections with amplitudes markedly larger or smaller than the Profile's are excluded. When the WeightingFlags mcpewfPointsForSameOffset bit is set, amplitude analysis proceeds as described in the above paragraphs. But then a further analysis is performed to adjust the weight given to a profile segment based on how closely its mean luminance value (i.e., the offset from zero luminance) matches the mean luminance value of the Pattern. This is the only case where the overall Pattern values matter; in all other cases only the shape of the pattern matters. The strength of the mcpewfPointsForSameOffset effect is inferred from the SizeWeighting property value and the mcpewfPointsForSameSize WeightingFlags bit. If the mcpewfPointsForSameSize flag is off, then the mcpewfPointsForSameOffset is actually stronger for smaller SizeWeighting property values (if you are looking for only big profile amplitudes, then offsets are likely less important and vice versa). If the mcpewfPointsForSameSize flag is on, then the mcpewfPointsForSameOffset is stronger for larger SizeWeighting values; that is, when both of these flags are on the SizeWeighting property indicates how closely the profile needs to exactly match the pattern for a high weight.