Geometric Distortion and TV Distortion

Geometric distortion is the relative difference between actual and ideal image height, caused by magnification changing with field position. It displaces geometry—bending lines or changing proportions—but is not the same as defocus blur. A common optical definition is D(%) = 100 × (Yactual−Yideal)/Yideal, where Yideal follows the selected projection model. Barrel distortion commonly bows edges outward; pincushion distortion pulls them inward. Confirm sign convention in the test specification.

For rectilinear projection, ideal image height is y=f tanθ. Fisheye and other wide-angle projections may instead use equidistant, equisolid-angle, or other mappings. The same real lens can therefore have different reported distortion percentages against different ideal models. State projection model, field point, wavelength, working distance, formula, and sign convention.

TV distortion is commonly measured from a rectangular chart as the image-border deviation from an ideal frame. One common convention is DTV = 100 × ΔH/H, where H is the specified image height and ΔH is maximum edge deviation; other specifications use conventions such as 100 × Δh/(2h). TV distortion emphasizes visible overall shape error, while optical distortion is often calculated from image-height-versus-field-angle data or grid mapping. The values are not interchangeable without matching target, boundary, and formula.

Distortion correction fits radial/tangential calibration terms and remaps pixel coordinates. Resampling can crop edges and reduce local effective resolution; model error and changes in temperature or focus can leave residuals. For localization, metrology, or multi-camera fusion, validate maximum residual after correction instead of relying only on a nominal distortion percentage.