detection of substation equipment can promptly and effectively discover equipment overheating defects and prevent equipment failures. Traditional manual diagnosis methods are difficult to deal with the massive infrare...
详细信息
detection of substation equipment can promptly and effectively discover equipment overheating defects and prevent equipment failures. Traditional manual diagnosis methods are difficult to deal with the massive infrared images generated by the autonomous inspection of substation robots and drones. At present, most of the infrared image defect recognition is based on traditional machine learning algorithms, with low recognition accuracy and poor generalization capability. Therefore, this paper develops a method for identifying infrared defects of substation equipment based on the improvement of traditional ones. First, based on the Faster RCNN, targetdetection is performed on 6 types of substation equipment including bushings, insulators, wires, voltage transformers, lightning rods, and circuit breakers to achieve precise positioning of the equipment. Afterwards, different classes are identified based on the sparse representation-based classification (SRC), so the actual label of the input sample can be obtained. Finally, based on the temperature threshold discriminant algorithm, defects are identified in the equipment area. The measured infrared images are used for experiments. The average detection accuracy achieved by the proposed method for the 6 types of equipment reaches 92.34%. The recognition rate of different types of equipment is 98.57%, and the defect recognition accuracy reaches 88.75%. The experimental results show the effectiveness and accuracy of the proposed method.
We are developing machine learning algorithms to identify chemicals of interest by their diffuse infrared reflectance signatures. The infrared Backscatter Imaging Spectroscopy (IBIS) technology is used to capture thes...
详细信息
We are developing machine learning algorithms to identify chemicals of interest by their diffuse infrared reflectance signatures. The infrared Backscatter Imaging Spectroscopy (IBIS) technology is used to capture these reflectance signatures through a mobile cart-based system. An infrared laser is directed toward the target surface from a standoff distance, and an MCT focal plane array collects the backscatter images from different wavelengths of laser into what is known as a hyperspectral image cube (hypercube for short). The hypercubes from our measurements are represented in a matrix of numbers. These hypercubes have a 128x128 front face representing the spatial dimensions (the physical image), and a depth of 153 representing the spectral dimension, which contains the IBIS measurements from various wavelengths of laser. These data are fed into a deep neural network machine learning algorithm for classification into specific chemicals (analytes). We propose a data accentuation technique to improve classification performance on IBIS data. Furthermore, we demonstrate the capability of machine learning algorithms to perform pixel-by-pixel discrimination of chemicals within a hypercube image, as well as the simultaneous detection of multiple different chemicals. We also demonstrate the capacity of machine learning algorithms to perform with high precision and recall under noisy conditions. The algorithm goes beyond simply identifying the presence of a chemical, in that it is also able to pinpoint the spatial location of each chemical within the field of view.
Aiming at the problem of high false alarm rate in the detection of dim infraredtargets in complex backgrounds, an infrared dim targetdetection method based on visual mechanism and clutter suppression model is propos...
详细信息
Aiming at the problem of high false alarm rate in the detection of dim infraredtargets in complex backgrounds, an infrared dim targetdetection method based on visual mechanism and clutter suppression model is proposed. Firstly, a weighted template is used to roam the image to enhance dim targets and suppress the background,then the image is divided into sub-regions, the local characteristics are used to suppress strong edge clutter. Finally the dim targets are selected according to the classification criteria. Experimental results show that the proposed algorithm can effectively reduce the false alarm rate in complex scenes, and signal-to-clutter ratio and background suppression factor are better than comparison algorithms.
Gastric cancer is one of the most serious cancer that affects and kills many people around the world every year. Early treatment of gastric cancer dramatically improves the survival rate. Endoscopy has become an impor...
详细信息
暂无评论