A low-illumination forest fire image segmentation method

Abstract

The invention discloses a low-illuminance forest fire image segmentation method, comprising the steps of: 1. importing the low-illuminance forest fire image collected by a camera into an image processor; The image is image enhanced; 3. The median filter is used to process the image; 4. The flame binarized image based on the YCbCr color space segmentation is obtained; 5. The flame binary image is obtained based on the region growing segmentation algorithm. Sixth, the image processor performs an XOR operation on the flame binarized image based on the YCbCr color space segmentation and the flame binarized image segmented based on the region growing segmentation algorithm to obtain the final low-light forest fire image segmented image. The method of the invention has simple steps, is convenient to implement, has high flame segmentation accuracy, high efficiency, good segmentation effect, strong anti-interference performance, strong practicability, good use effect, and is convenient for popularization and use.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a low-illuminance forest fire image segmentation method. Background technique [0002] In recent years, computer vision-based forest fire detection techniques have begun to replace traditional sensor-based forest fire detection methods. Image segmentation is the first and very important step in the application of computer vision technology. In the field of forest fire detection, many scholars have proposed a variety of algorithms for detecting fires in images or video sequences, such as the paper "Investigation The method proposed in "ofa novel image segmentation method dedicated to forest fire applications", Jiang Xiangang and others published the paper "Fire Region Based on HOFHOG and RDF" on pages 494-499 of the 2nd issue of "Computer Engineering and Design" in 2017 The method proposed in "Detection", and the paper "Flame Contour Extraction Algorithm Bas...

AI Technical Summary

Problems solved by technology

For example, the paper "Application of neural network to analyzes of CCDcolour TV-camera image for the detection of car fires" published by Ono T et al. In the method proposed in expressway tunnels, it is proposed to use the background image of the red component to extract the feature quantity of the potential area, and then input the feature quantity to the neural network model for training, and use the trained neural network to segment the flame pixels. It has high accuracy, but the neural network training process is more complicated, and the algorithm training time is longer; Celik et al. published papers on pages 176-185 of the Journal of Visual Communication & Image Representation, Issue 18, Volume 2, 2007 The flame segmentation method proposed in "Fire detection using statistical color model in video sequences" collects 150 fire pictures from the Internet, and generates a set of RGB color space rules, and then divides the flame pixels according to the decision rules. The method is good for flame segmentation in normal environment, but not ideal for flame segmentation in low-light environment; Chen Tianyan et al. in 2011, No. 30, Vol. 10, "Sensors and Microsystems", pages 62-64 The flame segmentation method proposed in the published paper "Flame Image Segmentation Method Based on YCbCr Color Space" proposes to analyze the distribution characteristics of flame pixels in YCbCr space, summarize the color decision of the flame, and finally segment the flame. The reporting rate is low, but the method introduces a changing threshold in the judgment condition, which has poor reliability and has high requirements for the segmentation environment, especially for the flame image segmentation effect under low illumination.

Images