Radial frequency (RF) contours can be used to represent the boundaries of natural objects, making them a useful stimulus with which to study shape processing. The current study used these contours to investigate whether luminance- and contrast-defined shape cues are combined in the detection of globally processed contours. A set of three experiments are presented. Experiment 1 shows that an RF contour defined by any one of positive or negative luminance contrast relative to background or second-order texture (contrast defined), is detected in accordance with a global pooling of local shape information. Experiment 2 uses a lateral masking paradigm to reveal interactions between global shapes defined by luminance of either polarity or a second-order contrast variation. Experiment 3 shows that an RF pattern defined by all three characteristics (positive, negative luminance polarity and contrast modulation) in nonoverlapping contour sections can still be assembled into a global shape. These findings indicate that at the level where RF contours are globally processed, luminance of either polarity- or contrast-defined shape information can be effectively combined.