Image analysis algorithms

For image analysis, algorithms will depend on the imaging solution used respectively. Various strategies have been employed to obtain consistent results, provided the actual number of cells analyzed per sample is sufficient (>3,000 cells). 

Curvature maps (Figure 4, left hand side) can be easily obtained by using image analysis algorithms previously developed. ... 

In order to quantify these observations, the pore size distribution was calculated using image analysis techniques. The used image analysis algorithms are described elsewhere [8] and enable the statistical distribution of the equivalent diameter of the pores Z)eq to be assessed. For the sake of comparison, only macropores smaller than 500 tun were evaluated, i.e. ultramacropores observed only in the Al foam were not considered. 

Furthermore, new image analysis algorithms have been developed based on perspective normalization and extraction of holistic features. These algorithms have been implemented and tested using footage obtained from CCTV cameras. Results have shown that these techniques can result in more than 90% accuracy when estimating crowd numbers. 

---