Most multirate image processing applications in multihop Wireless Mesh Network (WMN) are delay-sensitive but loss-tolerant. However, joint optimization of image quality and multihop Effective Capacity (EC) has been largely ignored in the current image transmission schemes in WMN. In this paper, we propose a new quality based cross-layer approach to adaptively optimize the image quality at the receiver side while satisfying the Quality of Service (QoS) requirements by exploring image position and value of magnitude (P-V) diversity. Specifically, the maximum multihop Effective Capacity (EC) is determined by QoS parameters and current link status of each path, and then the optimal source coding rate is selected by considering the position and value of magnitude information (F-V) of multirate image processing applications, which significantly increases expected image quality. Simulation results show that the proposed approach achieves better decoded image quality under the same QoS constraints.