Selective encryption for secure image transmission is an important ongoing research area which aims to scramble the critical information correlation and inter-dependency within final image compression bit streams. However, how to take Unequal Error Protection (UEP) method for transmitting partially encrypted image data in error-prone communication channels is still an open research problem in Wireless Sensor Networks (WSN). In this paper, we propose a new cross layer optimization approach for energy efficient delivery of selectively encrypted images in WSN while assuring image quality. In the proposed approach, identified importance levels of diversity for encrypted procession blocks as well as position-value (P-V) information are extensively explored within image compression and encryption bit streams. Desirable BER, ARQ retry limit, transmission rate are jointly optimized across PHY, MAC and APP layers regarding the image distortion reduction bound and communication energy efficiency. Both image quality and security are assured while energy consumption is minimized. Simulations results have shown significant gain in image security assurance and transmission quality achievement with energy efficient communications in WSN.