A simple algorithm for detection of significant vertices for polygonal approximation of chain-coded curves

A Heuristic Strategy for Sub-Optimal ThickEdged Polygonal Approximation of 2-D Planar Shape

“…In order to justify this trade off, Sarkar [29] combined these two measures as a ratio, producing a normalized figure of merit (FOM) which can be computed as ration of CR and ISE. Unfortunately, the FOM as proposed by Sarkar turns out unfit for comparing approximations with different number of points [30] and a parameterized version of weighted sum of squared error ( n k ISE WE CR  ) has been proposed in [31]. Carmona [32], [33] showed that the value k = 2 leads to the best performance.…”

“…Consistency: Consistency guarantees that if the number of DPs is plotted against the error measure, the error value should monotonically decrease with the increase in number of DPs. Most of the algorithms are found to be non-monotonic [30] which can pose a problem for a user as it is difficult to select an appropriate parameter if the effect of change in value is not predictable. Fig.…”

Edge detection in range images of piled box-like objects

“…Assuming N the number of points in a scan line, M the number of segments and d i the distance of the point at position i from the model approximating the point, we used the quantity defined in (4) to perform the comparison. The quantity F OM has been introduced in [13] and used among others in [12] for comparing segmentation results. Lower values of F OM correspond to better segmentations.…”

A new thresholding approach for automatic generation of polygonal approximations

“…Sarkar [36] proposed the to make the tradeoff between the compression ratio (CR) and the total distortion (ISE) caused [22]. Rosin [32] showed that the two terms CR and ISE used by F OM are not balanced, causing the measure to be biased towards approximations with lower ISE (which can be easily attained by increasing the number of detected dominant points).…”