基于色彩通道融合的火焰图像分割

Segmentation of Flame Image Using Multiple Color Channel Fusion

提出了基于色彩通道融合的回转窑火焰图像分割方法。由彩色火焰图像的红绿通道、红蓝通道和绿蓝通道构成三幅新的图像 ,分别从中提取训练样本集 ,对三个神经网络进行训练。神经网络收敛后 ,各自对相应的图像进行分割 ,会得到三种不同的结果。采用均值、中值、模糊逻辑和神经网络四种方法将其进行融合 ,会得到很高的分割准确率。实验结果表明该方法是可行的。

One approach to segmentation of flame image in kiln based on color channel fusion is proposed.Three new images are created by every two channels from red,green and blue channel of flame inage.Three training sample sets are extracted from the three images and used to train three BP neural networks.When the networks are converged,they are used to segment the corresponding image channels respectively.Then the obtained three different results are fused by four fusion algorithms,i.e.,mean,median,fuzzy logic and neural network.Thus the final segment results are with higher accuracy.The experimental results show that the proposed algorithm is feasible.