LLS-based consecutive line segments detection approach

A new linear least squares based consecutive line segments detection approach is proposed. Sub line segments are detected by traditional approach, and 3 aspects of traditional idea are improved. First, a new mergence rule based on linear least squares is proposed to improve traditional approach for some of the cases to be detected. Second, mergence confidence degree is defined to tell sub line segment to merge with whether it´s left neighbor or right neighbor. Third, a simplified way for reducing computation error approach is utilized to satisfy the requirement of both accuracy and rate for some of the cases to be detected. The presented approach is applied to detect consecutive line segments on images taken from injection molding parts time on line, satisfactory effect is received.