Application of back propagation neural network in the classification of high resolution remote sensing image: Take remote sensing image of beijing for instance

A Review on Back-Propagation Neural Networks in the Application of Remote Sensing Image Classification

“…Hence, the network is theoretically ready to be used in a particular classification application and produce a classified image [3]. Fig.…”

“…For classification problems, usually the number of the output neurons is related to the number of classes; however, the number of input layer neurons is normally related to the number of bands of the remote sensing image being classified. In contrast, for the hidden layer, the number of the neurons is not readily decided because there is no fixed formula or adopted guideline to determine this number [3]. Activation function identifies the neuron output in terms of the linear summation of its inputs.…”

“…Fig. 6 shows flowcharts of the two broad steps involved in the classification of the remote sensing images as adopted by Jiefeng et al [3].…”

“…An example of such a task is the classification of a high resolution satellite images. Jiefeng et al [3] used 0.01 and 0.8 for the learning rate and momentum parameters respectively in a classification of high resolution satellite image by BPNN [3]. These values depend on the ANN designs and application; thus, these values are adopted after an investigation stage for the training samples.…”

“…ANN (Artificial neural network) technology is applied extensively in the classification process of remotely sensed image data. This is due to the ability of ANNs to solve discontinuous and non-linear classification problems, in addition to their robustness, adaptability, and high accuracy [3]. Different kinds of ANNs are constructed with different network architectures; thus different learning algorithms are required [4].…”