Flame real-time detection method for inspection robot

Abstract

The invention provides a flame real-time detection method for an inspection robot, and the method comprises the steps of collecting image frame data of a moving vehicle-mounted camera, carrying out the preprocessing, evaluating the definition of an image, and applying a motion blur restoration algorithm when a condition is not met; utilizing a trained deep convolutional network model to extract depth features of the preprocessed image; constructing a flame detector based on the depth features and a support vector machine; and using the flame detector for carrying out target detection on the image in the discrimination stage. According to the flame image detection method provided by the invention, the visual flame detection problem can be effectively solved, and the high efficiency and safety of flame identification are improved. The characteristics calculated by the deep convolutional network are used for replacing traditional artificial experience characteristics to extract flame related information, and the method has the advantages of high precision, high reliability and high robustness, and is suitable for application requirements of real-time flame task monitoring of the inspection robot.

Description

technical field [0001] The invention relates to the technical fields of computer vision and pattern recognition, in particular to a real-time flame detection method for inspection robots. Background technique [0002] With the continuous progress of industrialization, the continuous liberation of productivity, and the gradual improvement of living standards, building facilities and materials are becoming more and more dense and diversified, resulting in more and more fire factors. The reliability and accuracy of fire safety systems are raised. a higher challenge. Fire is one of the most common and most common major disasters that threaten public safety and social development. Its occurrence is often accompanied by huge direct or indirect economic losses, and even threatens people's lives. According to statistics, there were more than 160,000 fire alarms received in my country in the first half of 2018 alone. In response to fire, ancient people have long summed up the exper...

AI Technical Summary

Problems solved by technology

Considering that the movement of automatic inspection robots increases the difficulty of fire danger detection, that is, the previous method of using motion information to extract flame areas with fixed cameras or using context information through background modeling cannot meet this specific requirement.

Images