Some region descriptors are easily obtained based on all pixels of the region, such as area, center of gravity, and gray scale.
1. Area: Counting based on the number of object elements (PS: diverse calculation methods)
2. Regional Center of Gravity
3. Common density features of region density: Maximum, minimum, median, average, variance, and high-order moment of gray scale (or color component. You can use the histogram to obtain more information.
Several Typical regional density special diagnostic descriptors:
(1) transmission rate (transmission): Proportion of the light penetrating the target to the Injected Light
T = light penetrating target/Incident Light
(2) optical density)
The ratio of the incident light to the light penetrating target (the reciprocal of the transmittance), and the base-10 logarithm is used.
OD = lg (1/T) =-LGT
The value range of the optical density ranges from 0 (100%) to infinity (no transmission)
(3) integrated optical density)
The sum of the pixel light density and the integral light density in the area are the weighted and above descriptors of the gray values in the histogram. The maximum value, minimum value, median, average value, variance, and higher moment can be obtained.
Ii. Topology DescriptorTopology studies the properties of graphs that are not affected by distortion deformation (excluding tearing or pasting,
Topological nature: Global, independent of distance
1. oulardb:The number of ORAS in a graph does not change with that in the graph. Therefore, you can use it to describe it.
2, the area of the moment: http://blog.csdn.net/dadaadao/article/details/6114989The moment-changing value does not change with the rotation, translation, mirroring, amplification, and reduction of the image.
From Weizhi note (wiz)
Target description (region-based description)
