Active thermography enables rapid, noncontact, single sided nondestructive evaluation (NDE) of metals and ceramics. However, because image formation depends on heat diffusion in a solid, thermography images suffer from blurring. Blurry images can be reconstructed with machine learning algorithms. Hopfield Neural Network (HNN) offers a promising approach to image recovery that does not require a large volume of training data. Quantum computing parallelism offers the possibility of achieving efficient image processing relative to classical counterparts. This study explores performance of Quantum Hopfield Neural Networks (QHNNs) in reconstructing blurry thermography images. Using experimental and synthetic thermography images, we show that QHNN running on a quantum computing simulator achieves higher reconstruction accuracy than HNN. The performance metrics are the Intersection over Union (IoU) and Structural Similarity Index Measure (SSIM) (0.9981 and 0.9975 respectively for QHNN, and 0.9870 and 0.9945, for HNN). These findings support the integration of quantum machine learning into active thermography NDE pipelines.