Incorporation of enhanced texture/transition modeling tools into a synthetic image generation model

The authors show that the incorporation of patterned spatial mixing between classes and within class texture can greatly increase the fidelity of synthetic image generation (SIG) for both visual assessment and quantitative analysis using multi/hyperspectral analysis tools. We have demonstrated how within and between classes, spatial/spectral variability can be introduced into a SIG model that includes a full radiometric processing chain. Further effort in this area needs to focus on exercising this capability for specific scenarios and evaluating the quantitative fidelity of the resulting images.