Concept

singular values — where it appears

The magnitudes a matrix stretches by along its own axes, whose smallest ones name the directions a fit cannot see. Their ratio is the conditioning of a fit, and the directions belonging to the smallest of them are the quantities the data does not contain.

Named by 12 essays across 6 fields — each of them below, with the objects they name alongside it.

× 4.6e+6seven flatand the rest stiffsingular value ÷ the largest, log scale, smallest firstσ₈/σ₇ = 4.6e+6168 parameters · 528 residuals

Seven numbers no picture can name

Shift a whole reconstruction by a metre and a half, turn it half a radian, scale it by 2.7, and every photograph of it stays where it was to a hundredth of a billionth of a pixel. Move one point by fifty millimetres and they move by two thirds of a pixel.

manyviews · Gauge
0.5120.1110100how finely each point is read off the picture (px, log scale)worst epipolar error against the exact geometry (px, log scale)raw pixelscentred and scaled1.1× apart at 0.25 px, 29.9× at 4 pxspread 102 against 2.5e+5

Eight points and the basis they are read in

The linear system that recovers a fundamental matrix is written in whatever coordinates the marks were read in, and pixel coordinates are a bad choice. Centring and scaling them first is worth nothing at a quarter-pixel reading and a factor of thirty at four.

twoviews · Eightpoint
centre of visioncorrect from 17 cm, at 160 mm wide50° across

Which way the drawn circle leans

Two rules are given for the direction a drawn circle's short axis runs in: along the axis of the cylinder, and pointing at the centre of vision. On the optical axis both are exactly right. At the edge of an ordinary frame the first is out by three and a half degrees and the second by seventy-nine. The statement neither of them is comes out of differencing the projection, and it matches the drawn ellipse to a hundredth of a degree.

wrong · Conic
left pictureright pictureevery one of them fits every mark1.6e-6 px

A flat scene fixes no second eye

Eight marks on one plane leave the eight-point design matrix two ranks short, so a two-parameter family of fundamental matrices satisfies every mark exactly — three of its members are 1.41 apart after normalisation and all three fit to 1.6 × 10⁻⁶ pixels. The same photographs determine the plane's own map from four marks, to 5.9 × 10⁻¹³ pixels.

twoviews · Planar
0306090-3-2-1out-of-plane spread, as a fraction of the scene's own extent (powers of ten)error in the recovered translation direction, in degreeswith 0.3 px of reading errorexact marksthe geometry is a step and the measurement is a slope9 reliefs

How flat is flat enough

With exact marks the transition has no width at all — 84° of pose error at exactly coplanar and 0.000° at eight parts in ten thousand of relief. Put three tenths of a pixel of reading error in and the same sweep becomes a slope three decades wide, crossing into usefulness when the out-of-plane parallax reaches about ten times the marking error.

twoviews · Planar
00.50011.50210203040how many marks the fit was givenerror in the recovered translation direction, in degrees (powers of ten)flatwith depththe same reading error in both5 counts

An ambiguity is not an uncertainty

Eight marks to forty cuts a solid scene's pose error from 19.8° to 0.7° and leaves a flat one at 48°. The two failures look identical from inside — a confident answer, a residual at the floor — and they respond to opposite remedies, so telling them apart is worth more than either measurement.

twoviews · Critical
024010203040how far the mirror is turned from square, in degreeswhat three tenths of a pixel can cost, in millimetresnearly squareone mark, 0.3 px of marking error3.0× across the range

Square to the camera is the worst mirror

A mirror pair's baseline runs along the mirror's normal, so a mirror facing the camera puts the second eye directly behind the first — the forward-motion arrangement, with the epipole in the middle of the frame and the rays to a mark crossing at 23°. Turning it forty-four degrees opens that to 65° and cuts the worst depth error threefold, and the number to watch is not the angle but where the reflected lens sits on the print.

mirrors · Mirrorpair
the face's normal, drawnratio 0.8112 against the template's 0.5774tilt 0°major axis ⟂ the drawn normal

A circle off the coordinate planes

An ellipse template is cut at one ratio, and the ratio is the cosine of one angle: between a coordinate plane's normal and the direction of projection. A face tilted out of that plane needs a different ratio and — the half that gets drawn wrong even when the ratio is close — a major axis pointing somewhere else, perpendicular to the drawn normal rather than to any edge.

parallel · Isocircle
horizon25withheldfive marks fitted, one withheldcorrect from 26 cm, at 160 mm widethe withheld mark is 2e-13 px off the fitted conic

Five marks and the sixth

Five points determine a conic exactly — five coefficients up to scale, five equations, nothing left over — so a fit through five marks on a photograph is not a fit at all. The sixth mark, withheld, lands on the curve to 1.9e-13 px. And the moment a sixth mark is used, the arithmetic changes character completely: it becomes a least-squares problem, and the residual starts telling you something the five could never say.

foundations · Fiveconic
horizonflat on the groundleaning 23.6°two circles, 6.4 m across, in planes 23.6° apartcorrect from 26 cm, at 160 mm widetwo poses, 23.6° apart, one picture

Two circles, one picture

A photographed circle leaves its own pose ambiguous, and not a little: two congruent circles in planes 23.6° apart draw the same conic to 1.1e-16 on normalised coefficients, both of them in front of the camera. On top of that the distance is free, so the family is two discrete poses each with one continuous parameter — and a plane one degree from either draws a conic 2.5e-4 away, which is what makes the agreement a measurement.

foundations · Fiveconic
face 15.75×1.00× parallelface 21.41×1.00× parallelface 31.34×1.00× parallelface 41.13×1.00× parallelface 51.05×1.00× parallelface 61.03×1.00× parallelface 71.00×1.00× parallelface 81.00×1.00× parallelthe spread of scale within one faceparallel: 1.000000×

One face, one scale

A design carried from a point onto a flat face varies in scale by nearly six across that one face; the same design carried along a direction varies by 1.000000000. The stretch is reported here as the two singular values of the local map rather than as one directional difference, which is the honest form and which the previous round owed.

parallel · Noeye
fitted on five lines, tested on seven points7.4e-10 px

Five tangents name the same conic

Fitting a conic to five lines is the same six-coefficient nullspace problem as fitting one to five points, with the roles exchanged, and it lands on the same curve to 2.6 × 10⁻¹¹ pixels. What duality does not promise is that the two are equally good evidence — and measured at equal marking precision they are, to within seventeen per cent, because a drawn tangent is made of points.

foundations · Fiveconic

Named alongside it

The objects these essays reach for when they reach for this one.

ConditioningConicHomographyJacobianreconstruction ambiguityDegeneracyDegenerate configurationFundamental matrixleast squaresAnisotropyBaselineCircle

All concepts