Low-illumination-level l forest fire image segmentation method

Abstract

The invention discloses a low-illumination-level forest fire image segmentation method which comprises the steps of (1) introducing a low-illumination-level forest fire image collected by a camera into an image processor, (2) adopting an improved histogram equalization algorithm to carry out image enhancement processing on the low-illumination-level forest fire image, (3) carrying out median filtering processing on the image by using a median filtering method, (4) obtaining a flame binary image based on YCbCr color space segmentation, (5) obtaining a flame binary image based on regional growthsegmentation algorithm, and (6) allowing the image processor to carry out XOR operations on the flame binary image based on the YCbCr color space segmentation and the flame binary image based on theregional growth segmentation algorithm, and obtaining a final low-illumination-level forest fire segmentation image. The method has the advantages of simple steps, convenient implementation, high accuracy of flame segmentation, high efficiency, a good segmentation effect, strong anti-interference performance, strong practicability and a 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 technology has 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 of a novel image segmentation method dedicated to forest fire applications", Jiang Xiangang et al published the paper "Fire Based on HOFHOG and RDF" on pages 494-499 of the 2nd issue of "Computer Engineering and Design" in 2017 The method proposed in "Regional Detection", and the paper "Flame Contour Extraction Algorithm Based on YIQ Color Space" publ...

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" published by Ono T et al. The method proposed in "fires in expressway tunnels" proposes 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. The method has high accuracy, but the neural network training process is more complicated, and the algorithm training time is longer; Celik et al. published on pages 176-185 of the 2007 18th volume 2 journal "Journal of Visual Communication & Image Representation" The flame segmentation method proposed in the paper "Fire detection using statistical color model in videosequences" collected 150 fire pictures from the Internet, and generated a set of RGB color space rules, and then segmented 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; 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