Study on two-dimensional distribution of X-ray image based on improved Elman algorithm

The principle of the X-ray detector which can simultaneously perform the measurement of the exposure rate and 2D (two-dimensional) distribution is described. A commercially available CMOS image sensor has been adopted as the key part to receive X-ray without any scintillators. The correlation between the pixel value (PV) and the absorbed exposure rate of X-ray is studied using the improved Elman neural network. Comparing the optimal adjustment process of the BP (Back Propagation) neural network and the improved Elman neural network, the neural network parameters are selected based on the fitting curve and the error curve. The experiments using the practical production data show that the proposed method achieves high accurate predictions to 10 −15 , which is consistent with the anticipated value. It is proven that it is possible to detect the exposure rate using the X-ray detector with the improved Elman algorithm for its advantages of fast converges and smooth error curve.