An Accurate and Fast Pattern Localization Algorithm for Automated Visual Inspection

A Hybrid Image Alignment System for Fast and Precise Pattern Localization

“…The energy function to be minimized in this algorithm is a sum of squared modified optical flow constraints [14] given as follows EðÁx; Áy; ; ; a; bÞ…”

“…The optical-flow based energy minimization algorithm [14] is an accurate, efficient and robust algorithm for determining precise pattern localization. This algorithm is based on a least squares fitting of the modified optical flow constraints.…”

“…This method is in principle a template matching-based approach, so it is also robust against noises. The FLASH algorithm consists of two stages; namely, a hierarchical nearest-neighbor (HNN) search procedure for approximate pattern localization and an optical-flow based energy minimization method [14] for accurate estimation of the pose. The hierarchical nearest-neighbor search procedure produces rough estimates of the transformation parameters as the initial guesses for the optical-flow based energy minimization procedure, which then iteratively refines the parameter estimation.…”

“…The proposed FLASH algorithm comprises a hierarchical nearest-neighbor (HNN) search procedure and an optical-flow based energy minimization procedure [14]. The flow chart of the FLASH algorithm is shown in Figure 2.…”

Robust Image Matching under Partial Occlusion and Spatially Varying Illumination Change

“…Some previous work [9,10] used a modified brightness model to account for uniform photometric variation; i.e.,…”