We explore the application of Bayesian image analysis to infer the properties of an SDSS early-type galaxy sample including AGN .
We use GALPHAT ( ) with a Bayes-factor model comparison to photometrically infer an AGN population and verify this using spectroscopic signatures .
Our combined posterior sample for the SDSS sample reveals distinct low and high concentration modes after the point-source flux is modeled .
This suggests that ETG parameters are intrinsically bimodal .
The bimodal signature was weak when analyzed by GALFIT ( ) .
This led us to create several ensembles of synthetic images to investigate the bias of inferred structural parameters and compare with GALFIT .
GALPHAT inferences are less biased , especially for high-concentration profiles : GALPHAT Sérsic index n , r _ { e } and MAG deviate from the true values by 6 \% , 7.6 \% and -0.03 \mathrm { mag } , respectively , while GALFIT deviates by 15 \% , 22 \% and -0.09 mag , respectively .
In addition , we explore the reliability for the photometric detection of AGN using Bayes factors .
For our SDSS sample with r _ { e } \geq 7.92 arcsec , we correctly identify central point sources with \mathrm { Mag _ { PS } } - \mathrm { Mag _ { Sersic } } \leq 5 for n \leq 6 and \mathrm { Mag _ { PS } } - \mathrm { Mag _ { Sersic } } \leq 3 for n > 6 .
The magnitude range increases and classification error decreases with increasing resolution , suggesting that this approach will excel for upcoming high-resolution surveys .
Future work will extend this to models that test hypotheses of galaxy evolution through the cosmic time .