Intelligent flame detection method based on image segmentation recognition

[0019] like figure 1 As shown, a flame intelligent detection method based on image segmentation recognition, including:

[0020] Collect real-time on-site image information T(T1(x,y), T2(x,y), T3(x,y), T4(x,y)), T1(x,y) refers to the color characteristics of the real-time flame, T2(x,y) refers to the motion characteristics of the real-time flame, T3(x,y) refers to the geometric characteristics of the real-time flame, and T4(x,y) refers to the texture characteristics of the real-time flame;

[0021] Call the K(K1(x,y),K2(x,y),K3(x,y),K4(x,y)) of the preset scene, K1(x,y) refers to the color characteristics of the preset flame , K2(x, y) refers to the movement characteristics of the preset flame, K3(x, y) refers to the geometric characteristics of the preset flame, K4(x, y) refers to the texture characteristics of the preset flame; the preset scene refers to The scene where there is no fire in normal operation;

[0022] Calculate correlation A=k 1 Q1+k 2 Q2+k 3 Q3+k 4 Q4,