Energy efficient and load-balanced routing based on QoS requirements in Wireless Multimedia Sensor Network have been achieved using Energy Enhanced Load Balancing Pairwise Directional Geographical Routing (EELB-PWDGR). However, the process of path selection is a time-consuming one. Hence, Artificial Bee Colony (ABC) is used to select the optimal path satisfying QoS constraints. Though the ABC-EELB-PWDGR outperforms EELB-PWDGR, the ABC algorithm exhibits unsatisfactory performance with a lower search speed, poor population diversity, stagnation within working methods, and struck to the local optimal solution. This study uses an enhanced ABC algorithm in which the global best solution information is added into the solution search equations in order to find the new solution only around the best solutions of the previous iterations for improved exploitation. Since Population initialization is imperative due to its impact on convergence speed, an initialization approach which based on chaotic systems and opposition based learning method has been employed to balance the diversity and convergence capability of ABC. Thus the Enhanced ABC based EELB-PWDGR (EABC- EELB-PWDGR) constructs an initial population with a maximum diversity to provide the best search solutions and a high degree of accuracy. The experimental results prove that the proposed EABC-EELB-PWDGR provides better routing performance than ABC-EELB-PWDGR.