Algorithm for digital location and recognition of digital instruments in complex scenes is proposed. First of all, the digital instrument images collected during the inspection process of the substation inspection robot are preprocessed by gray scale, filtering and morphology. Secondly, looking for extreme area, and using extreme values for coarse positioning, and then extract the ROI area of the number. In the ROI region, the preprocessing operation such as filtering and morphology are performed again, and the extreme region is searched, the error window is eliminated which can be achieve accurate positioning and segmentation of digital. Finally, the digital feature vectors are extracted and sent into the SVM classifier in sequence. The determination of significant figure of result based on the first nonzero value, and the correct reading of the meter is exported. The experimental results show that the algorithm can realize the digital location and recognition of digital instrument in complex scene. The algorithm has high accuracy and strong robustness, which can be met the requirement of digital instrument for substation inspection robot.
[1]
Qi Bing-juan,et al.
An Overview on Theory and Algorithm of Support Vector Machines
,
2011
.
[2]
Bo Li,et al.
A vehicle license plate recognition system based on analysis of maximally stable extremal regions
,
2012,
Proceedings of 2012 9th IEEE International Conference on Networking, Sensing and Control.
[3]
Lu Shouyin,et al.
Development of a Mobile Robot for Substation Equipment Inspection
,
2006
.
[4]
MA Yue-mei.
Textural image filtering method by hybridizing wave atoms shrinkage with bilateral filtering
,
2013
.
[5]
Yan Si-ha.
Feature line extraction method for heritage image
,
2013
.