研究目的
Investigating the extraction of multiple features from insulator images to provide information for insulator detection and recognition, thereby enhancing the safety of the electric power system.
研究成果
The study successfully extracts multiple features from insulator images using computer vision technology, providing valuable information for insulator detection and recognition. The methods employed, including texture analysis, invariant moment features, and geometric characteristics computation, demonstrate effectiveness in enhancing the safety management of insulators in power systems.
研究不足
Some images cannot be effectively separated by pixel statistics due to the complex environment of substations. The extraction algorithm of multiple eigenvalues has certain practicability but may not be universally applicable.
1：Experimental Design and Method Selection:
The study employs computer vision technology to extract characteristic values from insulator images. Methods include pixel statistical segmentation, Gray level co-occurrence matrix for texture features, invariant moment features extraction in binary images, and geometric features computation.
2：Sample Selection and Data Sources:
Actual insulator images collected from high voltage transmission lines in Shan-Xi province, with a resolution of 800×
3：List of Experimental Equipment and Materials:
6 Digital camera for image collection, Matlab R2016B software for simulation experiments.
4：Experimental Procedures and Operational Workflow:
Images are converted to grayscale, binarized using optimal threshold segmentation, and then processed for texture, invariant moment, and geometric features extraction. Local features detection is performed using MSER, Harris, and SURF methods.
5：Data Analysis Methods:
Texture features are analyzed using entropy, contrast, correlation, second-order moment, and inverse difference moment. Invariant moment features and geometric characteristics like area, perimeter, and barycenter are computed.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
