Image dehazing is an increasingly widespread approach to address the degradation of images of the natural environment by low-visibility weather, dust and other phenomena. Advances in autonomous systems and platforms have increased the need for low-complexity, high-performing dehazing techniques. However, while recent learning-based image dehazing approaches have significantly increased the dehazing performance, this has often been at the expense of complexity and hence the use of prior-based approaches persists, despite their lower performance. This paper addresses both these aspects and focuses on single image dehazing, the most practical class of techniques. A new Dark Channel Prior-based single image dehazing algorithm is presented that has an improved atmospheric light estimation method and a low-complexity morphological reconstruction. In addition, a novel, lightweight end-to-end network is proposed, that avoids information loss and significant computational effort by eliminating the pooling and fully connected layers. Qualitative and quantitative evaluations show that our proposed algorithms are competitive with, or outperform, state-of-the-art techniques with significantly lower complexity, demonstrating their suitability for use in resource-constrained platforms.