Background:
Unmanned aerial vehicle automatic fault identification of high voltage transmission equipment
has entered the stage of product development, in which image recognition technology is one of the key technologies.
There are often bird nests on the high voltage transmission tower, which have an impact on the transmission, so they
needs to be automatically detected.
Methods:
For bird's nest recognition, a novel algorithm is proposed. Firstly, the template image and auxiliary function are
used to construct the system, and the iterative point trajectory set, called feature set, is obtained by iteration; Then, the
target image is searched by blocks, and the image blocks are iterated with the same auxiliary function to construct the
iterative system, and the set of iterative point tracks to be identified is obtained. The correlation coefficient is calculated
by comparing the feature set with those to be recognized. And we can confirm whether the image block is a bird's nest
according to the size of correlation coefficient.
Results:
Different from the general image recognition method, the iterative algorithm obtains the iterative trajectory by
iterating the image and the auxiliary function, and takes the iterative trajectory as the image feature, then the feature
comparison is carried out, so as to achieve the goal of bird's nest recognition. The effectiveness of the method is proved by
experiments. The recognition accuracy is 99% by experiment on the self-built data set.
Conclusion:
This paper proposes a new feature extraction algorithm for bird's nest recognition. The algorithm based on
iteration is very simple and effective for bird's nest identification. As a new method, it needs further development and
improvement.