To overcome the limitation of the Grasselli's morphology parameter only represent the local features of the fracture roughness, this paper defines an average slope angle to reflect the ignored roughness information. A modified parameter \({\theta }_{C}\) is proposed by incorporating the average slope angle into the Grasselli's morphology parameter, and its ability to capture the anisotropic characteristics of joint morphology is validated. Direct shear tests are performed on joint replicas with different morphology to investigate the relationship between of the modified Grasselli's morphology parameter and shear strength. The results show the contribution of \({\theta }_{C}\) to peak dilation angle depends on the \({\sigma }_{\text{n}}/{\sigma }_{\text{c}}\) ratio. Follow the physical constraints, a peak dilation angle model is constructed. The initial dilation angle, as determined from tilt tests, could be expressed as twice the \({\theta }_{C}\). Finally, a new shear strength criterion for rock joints is proposed. Compared to existing criteria, this criterion has a simpler form and provides a more comprehensive understanding of dilation behavior. The predicted results indicate that it reliably estimates the joint shear strength.