Optimized Mixture Kernels Independent Component Analysis and Echo State Network for Flame Image Recognition

In this paper, an optimized mixture kernels independent component analysis and echo state network flame image recognition model is proposed. Firstly, in order to describe the flame image feature information in detail, 19 feature vectors of the three types of color, shape and texture feature are comprehensively extracted; a mixture kernels independent component analysis method is proposed to perform nonlinear transformation and reduce the correlation of them. The mixture kernels function among them is a combination of linear kernel function, Matern kernel function and Gaussian radial basis kernel function. Then, the feature vector after nonlinear transformation isused as the input vector, and the echo state network model is trained as the recognition model. At the same time, a normal cloud-black hole optimization algorithm (NCGBH) combining black hole algorithm with cloud model is proposed, which can optimize several adjustable parameters of the model. Four benchmark functions are firstly used for simulation experiments to prove advancement of the proposed NCGBH algorithm; and then, several flame images are used to verify performance of the proposed recognition model. From the comparison and analysis, it is verified that the proposed method has good results in recognition performance, generalization ability, and computational time.