Table 3.
The correlation and the distance between the manual result from observer A and the algorithmic results with standard parameters and after the parameters are changed by ±20 %
| Parameter | Performance TP [FP]* (%) | Distance Error (pixel) | ||||
|---|---|---|---|---|---|---|
| Change | 0 % | −20 % | +20 % | 0 % | −20 % | +20 % |
| Max shadow intensity threshold | 94 [4] | 91 [5] | 94 [7] | 1.7 ± 1.1 | 1.8 ± 1.3 | 1.8 ± 1.3 |
| Sliding shadow size | 94 [4] | 92 [5] | 93 [5] | 1.7 ± 1.1 | 1.7 ± 1.2 | 1.7 ± 1.2 |
| Slope threshold | 94 [4] | 93 [7] | 92 [5] | 1.7 ± 1.1 | 1.8 ± 1.3 | 1.8 ± 1.2 |
| Shadow edge length threshold | 94 [4] | 85 [12] | 88 [5] | 1.7 ± 1.1 | 1.8 ± 1.3 | 1.8 ± 1.3 |
| Clustering distance threshold | 94 [4] | 94 [6] | 93 [4] | 1.7 ± 1.1 | 1.8 ± 1.2 | 1.7 ± 1.2 |
The distance error and its standard deviation are measured in pixel size
* Performance TP value means the sensitivity our method. FP value in [] shows the percentage of false positive detected struts compared to the manual results