Patterns tracking using local lacunarity and track-before-detect approach

Line tracking is important for numerous applications of the machine vision, including control of line following robots. Local lacunarity for binary images allows the estimation of line by the detection of spatial differences using statistical properties of textures. Additional tracking algorithm improves the line detection by application of motion model. Such approach allows the extended width line tracking. The properties of proposed local lacunarity are analyzed for the perturbed pattern located on the random background. Monte Carlo approach is applied for the analysis of such lines for two smoothing coefficient values of Spatio-Temporal Track-Before-Detect algorithm. Obtained horizontal mean error value is comparable to the width of the line.