In depth from focus methods, a crucial step is to accurately estimate the sharpness or focus level at each image pixel using a focus measure (FM). However, despite the numerous FM proposals in literature, no single FM has proven effective across all scenarios. The derivative-based domain of FMs has primarily focused on first and second-order derivatives, which have been used to assess image sharpness. In contrast, this paper investigates higher-order directional derivatives (HODDs) for depth from focus methods. By exploring different approximations to HODDs, the impact of these on depth estimation quality is investigated. The responses from various order derivatives are then combined to produce an enhanced focus volume. The weights used to combine FMs are calculated based on their agreement with maximum focus value frame selection and kurtosis values. A series of experiments have demonstrated that the proposed HODD-based FMs are highly effective in assessing image sharpness. The proposed approach has been applied to benchmark datasets consisting of both synthetic and real focal stacks. The results show that the proposed method outperforms other approaches in providing higher-quality depth maps.