This paper combines Hopfield neural network with mathematical morphology to detect wood defects edges in X-ray wood image. At first, A Hopfield neural network with an improved energy function was presented for edge detection of log digital images for the first step. Different from the traditional methods, the edge detection problem in this paper was formulated as an optimization process that sought the edge points to minimize an energy function. The dynamics of Hopfield neural networks were applied to solve the optimization problem. An initial edge was first estimated by the method of traditional edge algorithm such as Canny Algorithm. The gray value of image pixel was described as the neuron state of Hopfield neural network. The state updated till the energy function touch the minimum value. The novel energy function ensured that the network converged and reached a near-optimal solution. In the second step, a mathematical morphology method was used to eliminate the noises of the image. Taking advantage of the collective computational ability and energy convergence capability of the Hopfield network and the excellent denoising capability of mathematical morphology, the noises will be effectively removed. Comparing with the traditional methods, our method can obtain a vivid and noiseless defect edge in the experimental results.