This paper presents algorithms for estimating parameters that characterize weak levels of a printer artifact referred to as banding. Flat field test images are typically used as test patterns for banding evaluation; however, the images of this study contain complex image content to demonstrate the algorithm’s robustness and extend the utility of the defect characterization methods. The test images are from color printers in the development phase and include multiple visible defects such as banding, grain, and streaking. The banding characterization includes an estimation of the fundamental frequency and average power extracted from local regions dominated by low frequency content where banding is likely to be most visible and offensive. Grain and mottle defects combined with other image content form a difficult noise environment from which the quasi-periodic banding characteristics must be extracted. The algorithm is based on the autocorrelation function and uses special averaging and a pre-whitening filter designed to minimize the influence of the interfering factor. Experimental results show that this method provides accurate banding frequency and power characterization even for multiple banding sequences that are present in the image test area. This new algorithm proves computationally efficient and more accurate than parameter estimates based on frequency domain analysis using the power spectrum. Experimental results show accurate banding characterizations for periods ranging between 0.93 and 10.5 mm over a range of banding-tonoise ratios from 5.5 to -6.5 dB.