Automatic reading method for pointer dial scales

Abstract

The invention discloses an automatic reading method of pointer dial scales, belongs to the field of image recognition, and discloses an automatic reading method of pointer dial scales. S00, data acquisition: taking pictures shot by a robot as a test set, and taking manually shot images as a training set; S10, data labeling: labeling a pixel region by taking the pointer type instrument scale regionand the pointer region as a whole; and S20, designing an end-to-end pixel identification network structure, wherein the end-to-end pixel identification network structure comprises pixel identification of a pointer type instrument scale area and a pointer area. The relative relation between dial scales and pointers is considered, the problems of false detection and missing detection of a deep network are greatly solved, instability caused by influences of noise light distortion and the like in a traditional algorithm is avoided, and robustness and use convenience are achieved in use.

Description

technical field [0001] The invention relates to the field of image recognition, in particular to an automatic reading method for dial scales of pointers. Background technique [0002] At this stage, the recognition methods of pointer meters in substations are divided into traditional image recognition methods and recognition methods based on deep learning. Traditional image recognition methods are divided into Hough detection method, template matching method, maximum connected domain method, silhouette method, least squares fitting method, line drawing method and region growing method, etc. The method based on deep learning has the advantages of robustness, accuracy, and simplicity, and has become a research hotspot in recent years. Usually, the deep learning framework is used to locate the rectangular frame of the pointer instrument, and then the traditional detection method is used to detect the scale and pointer in the rectangular frame, and then the reading is obtained....

AI Technical Summary

Problems solved by technology

Among the traditional methods, the commonly used Hough detection method and the maximum connected domain method require different algorithm designs for different meters, which is not universal; the template matching method is cumbersome to operate in the practical process, and needs to be done in advance for each meter. Good template and characterize the scale line and the center point of the instrument; the least squares fitting method, the line drawing method is not robust in the image with noise

In addition, the traditional method is cumbersome to design for instruments with upper and lower pointer meter areas.

[0003] Among the methods based on deep learning, end-to-end deep learning methods are difficult to guarantee robustness in practical applications due to the black-box effect of deep networks.

For the recognition of scale points and pointers, because the scale points are marked, that is, point features are extracted, the algorithm lacks robustness in comparison.

Images