From: Feature analysis for classification of trace fluorescent labeled protein crystallization images
Feature | Symbol | Description | Formulation |
---|---|---|---|
*Edge [44] | η | No of graphs (connected edges) | η=|S| |
η 1 | No of graphs with a single edge | η 1=|S i |, where |L(S i )|=1 | |
η 2 | No of graphs with 2 edges | η 2=|S i |, where |L(S i )|=2 | |
η c | No of graphs whose edges form a cycle | η c =|S i |,where S i is a cyclic graph | |
η p | No of line normals | \(\eta _{p}=\sum \! \perp \! \left (S_{k} \right),\! \perp \! \left (S_{k} \right)\! =\! \left \{\!\! \begin {array}{lc} 1 & \exists l_{i}\in L_{k}\text {and }\exists l_{j}\in L_{k}\text {and} \\ & 70\leq \alpha \left (l_{i},l_{j} \right)\leq 90 \\ 0&otherwise \end {array}\right.\) | |
μ l | Average length of edges in all segments | \(\mu _{l}=\frac {{\sum \nolimits }_{i\in \textbf {L}}l_{i}}{\left |\textbf {L}\right |}\) | |
S l | Sum of lengths of all edges | \(S_{l}={\sum \nolimits }_{i\in \textbf {L}}l_{i}\) | |
l max | Maximum length of an edge | l max = max1≤i≤|L|(l i ) | |
c o | 1 if η c >0, 0 otherwise | c o =∃S,Sis a cyclic graph | |
l o | 1 if η p >0, 0 otherwise | l o =(∃l i ∈L k and ∃l j ∈L k and70≤α(l i ,l j )≤90) | |
η hc | No of Harris corners | [48] | |
*Hough | η hl | No of Hough lines | [49] |
μ hl | Average length of Hough lines | [49] |