摘要：Apply the separation principle of the physical contradictions in TRIZ into the simulation of forest fire smoke and propose a forest fire smoke recognition algorithm based on gray bit-plane(short in FFSGBP) to change the problem of identifying the smoke into the analysis on the bit-plane images. To get eight images of moving smoke, the separation process on smoke moving images is necessary; use high-diagram image (the 7th,the 8th image) to extract contours, and extract the contour of the remaining images in the same position. Superimpose the information extracted, get the smoke model. By adding motion operators to the original image, we could simulate the motion of discrete smoke better, by using mean filter, we could reduce the influence of noise, and it can improve the reliability of the experiments. The results show that compared with the traditional method based on HSI model, the proposed algorithm can achieve the fire smoke detection quickly and obtain the contour model of the smoke inside efficiently and accurately.
关键词：TRIZ theory; gray bit plane; contour extraction; mean filtering
