Detecting image points of general imbalance

Imbalance oriented selection scheme was recently proposed to detect stable image points in weakly or sparsely textured images. The scheme chooses image points whose one-pixel-wide directional intensity variations can be clustered into two imbalanced classes as candidates. In this paper, we propose general imbalance decided by multi-pixel-wide directional intensity variations. We present a case study of general imbalanced points in road sign images, which demonstrates the good potential of general imbalanced points.