Fig. 1
Example of Usage. On the left, the original image is shown, while on the right SURF detections are represented as rectangles of different colours.
Table 1
Efficiency Results for mahotas, pymorph, scikits-image, and openCV (through Python wrappers). Shown are values as multiples of the time that numpy.max(image) takes to compute the maximum pixel value in the image (all operations are over the same image). For scikits-image, features on the grey-scale cooccurrence matrix were used instead of Haralick features, which it does not support. In the case of median filter, the radius of the structuring element is shown in parentheses. NA stands for “Not Available.”
| Operation | mahotas | pymorph | scikits-image | OpenCV |
|---|---|---|---|---|
| erode | 1.6 | 15.1 | 7.4 | 0.4 |
| dilate | 1.5 | 9.1 | 7.3 | 0.4 |
| open | 3.2 | 24.3 | 14.8 | NA |
| median filter (2) | 226.9 | NA | 2034.0 | NA |
| median filter (10) | 2810.9 | NA | 1877.1 | NA |
| center mass | 5.0 | NA | 3611.2 | NA |
| sobel | 34.1 | NA | 62.5 | 6.2 |
| cwatershed | 174.8 | 58440.3 | 287.3 | 44.9 |
| daubechies | 18.8 | NA | NA | NA |
| haralick | 233.1 | NA | 7760.7 | NA |