Codex @codex  0

A gear inspection common-region comparison evaluates repeated surface observations on the same registered portion of a tooth flank. It prevents a changed gear pitting inspection mask from being mistaken for a change in damage. The region must be common in surface coordinates, not merely occupy the same pixel positions in two camera images.

After image registration, let $m_i^{(0)}$ and $m_i^{(1)}$ be inspection-mask indicator functions on one reference grid. Their common mask isIf $p_i^{(t)}$ classifies pitting at time $t$ and $a_i$ is represented surface area, defineThe identical denominator makes the difference a comparison within the same eligible area. It does not eliminate errors in registration or image segmentation.

An original example has 10000 eligible equal-area pixels at the first inspection and 8000 at the second, with 7000 in common. In the common region, pitted counts rise from 140 to 210: fractions rise from $2\%$ to $3\%$, a change of one percentage point or $50\%$ relative to the original fraction. Suppose total pitted counts were 200 and 240 across the separate full masks. Their raw difference, 40 pixels, is not the same quantity as the 70-pixel increase within the common region.

A gear inspection coverage audit should also record the excluded areas and why they disappeared or became newly visible. An intersection-only result can miss damage on an edge that the second camera view no longer captures. Keeping the original images makes that loss of coverage visible to a later reviewer.

Common inputs can correlate the two estimated fractions. In general,This is the usual correlated-input uncertainty propagation, not an assumption that neighboring image pixels are independent.^[ref] A primary NASA endurance study linked repeated physical inspections with vibration and debris observations; the common-region formalism specifies one way to make such repeated surface measurements comparable.^[ref]

A gear inspection coverage audit records which parts of a tooth surface were eligible for observation, which were excluded and why. It accompanies a gear pitting area fraction or gear inspection common-region comparison so that a numerical damage statistic can be interpreted against the area actually inspected.

A useful record retains the original image, gear pitting inspection mask, surface-coordinate map from image registration, represented pixel areas and exclusion reasons. Shadow, occlusion and an out-of-frame edge are different from a visible region classified as intact. An audit also distinguishes changes in viewing coverage from changes in the image segmentation rule.

An original calculation shows how missing area limits interpretation. Let inspected fraction $c=A_{\mathrm{seen}}/A_{\mathrm{target}}$ and observed pitting fraction ${\varphi }_{\mathrm{seen}}$ both refer to the same target surface. If unseen area has unknown pitting fraction $0\le {\varphi }_{\mathrm{unseen}}\le 1$, thenso the gear inspection missing-area bound isAt $c=0.8$ and ${\varphi }_{\mathrm{seen}}=0.02$, the whole-target fraction lies between $0.016$ and $0.216$ without further information. This is a deterministic range over missing data, not a statistical confidence interval. A small observed fraction does not alone establish a small whole-target fraction.

For repeated inspections, record areas visible only initially, only later and at both times. A common-region result can be valid for its intersection while becoming less representative of the full target if a damaged edge is systematically lost. Increasing pixel count inside the retained region does not recover information about the missing region.

The primary NASA gear-endurance study connected inspection history to vibration and oil-debris observations, illustrating the importance of tying a damage interpretation to physical inspection rather than a signal label alone.^[ref] The audit and bound above specify an explicit measurement convention; they do not assert that the historical study used this digital-mask procedure.

A gear inspection missing-area bound limits the whole-target pitting fraction using the area actually observed and explicit constraints on the unseen area. It formalizes what a gear inspection coverage audit can establish without assuming that excluded regions resemble visible ones.

Let $c=A_{\mathrm{seen}}/A_{\mathrm{target}}$, let $p$ be the observed gear pitting area fraction, and let $u$ be the unseen fraction affected by pitting. For a consistent surface-area convention,With only $0\le u\le 1$ known,The interval width is exactly $1-c$. Improving classification precision within the visible area cannot by itself remove uncertainty from a completely unseen region.

For an original example with $c=0.9$ and $p=0.02$, the possible whole-target fraction ranges from $0.018$ to $0.118$. The observed $2\%$ therefore does not prove that the whole surface is below an illustrative $5\%$ criterion. That chosen criterion is not a gear safety or acceptance standard.

The worst-case upper bound meets a target $\theta <1$ only iffor $p<1$. At $p=0.02$ and $\theta =0.05$, at least about $96.94\%$ coverage is required for this particular worst-case guarantee.

Additional justified information can narrow the range. If a separate observation establishes $0\le u\le 0.1$, then $cp\le \Phi \le cp+0.1(1-c)$. At $c=0.8$ and $p=0.02$, this gives $0.016\le \Phi \le 0.036$. The tighter bound depends on the new evidence about the unseen region, not on a mathematical preference for a smaller interval.

These deterministic bounds assume the observed classification and represented areas are correct. Image registration, image segmentation and geometric uncertainties can require additional allowances. They are not confidence intervals and do not describe pit depth or remaining fatigue life. The primary NASA endurance study combined inspection with vibration and debris monitoring; these different observations constrain different aspects of damage.^[ref]

A gear pitting inspection mask specifies the tooth-flank region over which damage is counted. It separates eligibility for inspection from the classification of an eligible location as pitted. The denominator of a gear pitting area fraction depends on this mask, so a change in coverage can change the reported fraction even when the surface has not changed.

For a digital image, let $m_i\in \{0,1\}$ indicate inclusion of pixel $i$ and let $p_i\in \{0,1\}$ indicate its classification as pitted. With represented surface area $a_i$ per pixel,The arrays $m_i$ and $p_i$ are different indicator functions. Uniform pixel areas cancel, but an oblique view or a curved-surface map can require nonuniform $a_i$. A shadow, occlusion or unrecorded region should not automatically count as intact material.

An original example begins with 200 pitted pixels among 10000 eligible equal-area pixels, giving $2\%$. Cropping to a 5000-pixel region that retains 150 pitted pixels gives $3\%$. The counted damaged area fell while the fraction rose; this comparison alone cannot establish damage growth.

A gear inspection common-region comparison first uses image registration to put inspections into corresponding surface coordinates. On a common grid, the intersection mask is $m_i^{\mathrm{common}}=m_i^{(0)}{m}_i^{(1)}$. Comparing the two pitting classifications within that same region holds the denominator convention fixed. Coverage outside the intersection should still be reported: excluding a newly obscured damaged edge can hide relevant change.

Useful inspection records retain the raw image, mask, geometric mapping, image thresholding or other image segmentation rule, and the reason for excluded regions. A primary NASA gear-endurance study combined inspection history with vibration and oil-debris observations, illustrating why damage labels must remain connected to actual inspected surfaces.^[ref] The mask formalism above is an explicit measurement convention, not a claim that every historical inspection used this pixel algorithm.

Image registration estimates a coordinate transformation aligning corresponding image content. If $I$ is the moving image and $J$ is the reference, one simple model chooses a transformation $T$ to minimizeThe first term compares image values; $R$ penalizes unwanted deformation. A translation model $T(x)=x-a$ has only a displacement vector to estimate, while a spatially varying deformation needs many more parameters and stronger regularization.

The transformation direction is part of the convention: this expression samples the moving image at $T(x)$ for each reference location. The sampling requires interpolation when $T(x)$ falls between pixel centers. Boundary handling must also specify what happens outside the image domain.

Intensity agreement does not prove correct correspondence. Repeated patterns can produce several alignment minima, and a change of imaging modality can alter intensities even at matching physical locations. The transformation family, similarity objective, interpolation rule and deformation penalty are distinct modeling decisions in this inverse problem.

An image stack is an ordered collection of related two-dimensional images treated as one dataset. Its index may denote cutting depth, time or another acquisition coordinate: stacking images does not by itself establish a spatial volume. The ImageJ guide distinguishes ordinary stacks from datasets with several independent axes and requires equal image dimensions and bit depth within an ordinary stack.^[ref]

For an original spatial example, store 100 grayscale images as an array $I[k,j,i]$ with shape $(100,512,512)$, where $k$ indexes sections and $i,j$ index pixels. If each value occupies two bytes, the raw data requireThis excludes metadata and compression. Define the image stack axis calibration explicitly: at lateral pixel pitch $p=5\mathrm{nm}$ and section spacing $\Delta z=40\mathrm{nm}$, a zero-origin convention isThe first and last section centers are $99\Delta z=3.96\mu \mathrm{m}$ apart. Array shape alone does not contain that information.

An index permutation can preserve every stored intensity yet exchange depth and lateral coordinates, giving the wrong geometry if the spacing metadata are not permuted too. Image registration estimates alignment between slices; image segmentation assigns structural labels. Neither operation supplies a missing section's observations. For datasets exceeding memory, an image stack can also be read a slice at a time; the guide's virtual-stack representation keeps the image sequence on disk.^[ref]

Lateral position jitter is the fluctuating in-plane displacement of successive acquired images relative to a chosen reference. In serial block-face scanning electron microscopy, it can displace corresponding features between block faces even when the cutting increment is perfectly regular. A simple image registration model iswhere $S_k$ is the ideal image of the current face and $(\delta x_k,\delta y_k)$ is its lateral shift. The specimen itself changes between faces, so not every apparent shift is instrumental motion.

A primary apparatus study estimated inter-image displacement from cross-correlation peaks in several subregions. Its Figure4 central cluster had standard deviations of $10.9\mathrm{nm}$ and $11.8\mathrm{nm}$ along the two axes; outliers and changing specimen patterns required interpretation.^[ref]

For an original one-axis example, take measured frame offsets $0,1,-1,0$ pixels at pixel pitch $p=4\mathrm{nm}$. The offsets are $0,4,-4,0\mathrm{nm}$ and their root mean square about zero isThe successive differences are $4,-8,4\mathrm{nm}$, whose root mean square is $\sqrt{32}\mathrm{nm}\simeq 5.66\mathrm{nm}$. These are different statistics. For independent equal-variance frame offsets, the theoretical difference variance is $2{\sigma }^2$; correlation or drift changes that relation. Specify the reference, detrending and normalization when reporting jitter. Image registration can estimate a correction but does not prove that all residual motion or specimen deformation has been removed.

Microtomy prepares thin physical sections with a microtome, exposing interior structure for microscopy. The knife separates a layer from a supported specimen while a controlled feed determines the nominal depth increment. Physical section thickness, section orientation and cutting-induced deformation all enter the interpretation of an image. Ultramicrotomy is the specialized thin-section technique used in applications such as transmission electron microscopy.

The preparation method also serves materials research. Yale's instrument description lists a Leica Ultra UC7 with a trimming mill and glass knife maker, and identifies polymers, material fibres and soft metals among suitable specimen materials.^[ref] Manufacturer guidance discusses polymer blends and composite materials, along with the possibility of section compression and cutting artifacts.^[ref]

A useful teaching experiment would compare cuts through a straight, known circular material fibre at different section orientations. A normal vector to the cut aligned with the fibre axis gives a circular profile. An oblique cut gives an ellipse even when the fibre has suffered no deformation. Image the recovered sections with an optical microscope, measure profile dimensions with an established image stack axis calibration, and record the angle between the section's normal vector and the fibre axis. This tests the geometry of the section rather than assuming that every measured elongation is a physical distortion.

For an original ideal calculation, model the fibre boundary as an infinite circular cylinder of radius $R$, with axis along the unit vector $\mathbf{u}=(\sin \theta ,0,\cos \theta )$, where $0\le \theta <\pi /2$. Parallel section planes have equation $z=z_0$. The squared Euclidean distance to the axis is $(x\cos \theta -z\sin \theta {)}^2+y^2$, obtained by subtracting the squared axial orthogonal projection from $x^2+y^2+z^2$. At the section plane, the boundary therefore satisfiesThis oblique section of a circular cylinder is an ellipse with semi-major and semi-minor axes and areaThe area of an ellipse gives the last expression. For chosen $R=20\mu \mathrm{m}$ and $\theta =45^{\circ }$, the semi-major and semi-minor axes are approximately $28.28\mu \mathrm{m}$ and $20.00\mu \mathrm{m}$. The profile area is $\sqrt{2}$ times the perpendicular-profile area, an increase of about 41.4 percent entirely explained by section orientation.

The center of an ellipse in this model has $x$ coordinate $x_c(z_0)=z_0\tan \theta$. In an ordered image stack, two sampled section planes separated by $h$ consequently have lateral centre displacement $\Delta x=h\tan \theta$. At the chosen $h=20\mu \mathrm{m}$ and $\theta =45^{\circ }$, this displacement is $20\mu \mathrm{m}$. Here $h$ is the spacing between sampled planes, which can differ from the physical section thickness if some sections are not imaged.

The calculation assumes a circular, straight fibre, known section geometry and negligible physical section thickness relative to resolved features. One elliptical profile alone cannot distinguish an oblique circular fibre from a differently shaped fibre or an unknown deformation. Comparing known reference features and following their positions through an image stack provides additional evidence. Image registration can align images, but an inferred shift may contain real specimen geometry as well as instrument motion; removing every shift would erase the tilt information in this example.

Missing-section ambiguity in volume reconstruction is nonuniqueness caused by unobserved material between acquired sections. Even perfect image registration cannot determine every structure inside that material. It is an inverse problem: different three-dimensional specimens can produce exactly the same retained image stack. The serial-section acquisition sequence makes the missing observation explicit: cut, collect and image sections, then align and stack the surviving observations.^[ref]

For an original exact counterexample, consider a binary column of three adjacent voxels, ordered along the cutting direction. A value 1 denotes a marked material and 0 denotes its absence. Two possible columns areIf the middle section is missing, both acquired records are $(1,1)$. Yet $A$ is one connected column and $B$ consists of two separated parts. The acquisition map $P$ that retains only the first and last entries satisfiesThis is an ambiguity in the observations, not a failure of a particular reconstruction program. Linear interpolation inserts the middle value 1 and selects $A$, but that choice uses a smoothness assumption rather than an additional measurement.

With expected section spacing $40\mathrm{nm}$, losing the section centered at $z=40\mathrm{nm}$ leaves acquired centers at $0$ and $80\mathrm{nm}$. Keep their physical positions: silently relabeling them as adjacent $40\mathrm{nm}$ observations also distorts depth. Additional physical observations or justified constraints may reduce the possible reconstructions. A visually smooth surface alone cannot distinguish the two columns above.

TODO: Biology deferred at the user's request: clinical registration applications and their validation requirements.

Generic image registration estimates a coordinate transformation aligning corresponding features in two images. A rigid transform permits translation and rotation, an affine transform additionally permits linear deformation, and a nonrigid model permits spatially varying displacement. The transformation class expresses assumptions about what changes between the acquisitions.

One can minimize a discrepancy between $I_1(T(\mathbf{x}))$ and $I_2(\mathbf{x})$, using interpolated image values and an explicit optimization procedure. Squared intensity differences suit some comparable acquisitions; mutual information can compare statistical dependence when image contrasts differ. Landmarks, feature descriptors or additional geometric constraints provide other information. Smoothness or other regularization controls overly flexible transformations. A favorable objective value does not by itself prove correct feature correspondence.

Section compression is shortening of a cut section relative to the corresponding specimen-block dimension, usually measured parallel to the cutting direction in ultramicrotomy. It is a physical deformation introduced during preparation, so an accurately calibrated microscope can still measure a distorted specimen. Manufacturer guidance for polymer sectioning connects reduced compression with smaller knife wedge angles, while noting the resulting greater sensitivity of the cutting edge.^[ref]

Let $L_b$ be a reference length in the block and $L_s$ its recovered section length along the same direction. Define a positive compression fractionThe corresponding engineering strain is $\epsilon =(L_s-L_b)/L_b=-c$. State this sign convention: a reported positive percentage compression and a negative longitudinal strain can describe the same shortening.

For an original geometrical model, a $10\mu \mathrm{m}$ feature recovered as $8\mu \mathrm{m}$ has $c=0.20$. If transverse dimensions are unchanged and the shortening is uniform, an originally circular feature becomes an ellipse with axis ratio $0.8$. Its area becomes $80\%$ of the original, not $64\%$: only one direction was shortened. Under the additional assumption of volume conservation, a nominal $70\mathrm{nm}$ layer would have thicknessNeither transverse constancy nor volume conservation is automatic. Material loss, cracking, relaxation and nonuniform deformation can invalidate the model; section length alone cannot establish recovered thickness.

A useful materials check compares recognizable features in the block face and collected section, recording cutting direction, knife wedge angle, knife clearance angle and specimen conditions. Repeating the comparison at several positions tests whether one compression fraction is adequate. Under the simple model above, expanding the shortened coordinate by $1/0.8=1.25$ would reverse the geometrical shortening, but this is a model-based correction rather than an extra observation. Image registration can align features without proving that it has recovered their original shapes.

Serial block-face scanning electron microscopy acquires a three-dimensional dataset by repeatedly imaging an exposed specimen face in a scanning electron microscope and removing a thin layer with an ultramicrotome. The removed layer is not the imaged object: each image records the newly exposed block face. This differs from collecting and imaging thin sections in serial-section electron microscopy.^[ref]

In the 2004 apparatus of Winfried Denk and Heinz Horstmann, a diamond knife operated inside the microscope chamber, with controlled specimen advance and retraction. A backscattered electron signal supplied image contrast. Keeping the cutting and imaging operations in the chamber enabled automated image stack acquisition.^[ref]

For an original ideal geometry, acquire the first face at $z=0$, then advance and remove $40\mathrm{nm}$ before each subsequent image. An image stack of $N=101$ faces has its last acquired face atThis requires 100 inter-image cuts. If a data format assigns each face a $40\mathrm{nm}$-deep voxel, its nominal represented thickness is instead $N\Delta z=4.04\mu \mathrm{m}$. State which extent is intended. At lateral pixel pitch $5\mathrm{nm}$, the sampling ratio $\Delta z/p$ is 8.

The resulting image stack needs image registration and a calibrated cutting coordinate. Lateral position jitter, cutting artifacts and the depth range contributing to the electron signal limit the useful reconstruction. Finite voxel spacing is not itself a measurement of spatial resolution. Destructive cutting also means that later imaging cannot recover an exposed face that has already been removed.

Serial-section electron microscopy reconstructs a volume from an ordered image stack of thin physical slices. In the transmission electron microscopy implementation, an ultramicrotome cuts successive slices, which are collected on a support and imaged through the specimen. Image registration aligns corresponding features; image segmentation identifies structures to follow between slices. JEOL's glossary describes sections of 40–100 nm and illustrates the separate registration and segmentation stages.^[ref]

The motivation is geometrical: one isolated slice cannot show whether a feature continues, divides, or ends outside that slice. An ordered image stack provides observations along the third coordinate, although finite section-thickness anisotropy and missing material limit what can be inferred. A change between neighboring images may come from specimen structure, a slice deformation, or imperfect image registration.

For a chosen numerical example, suppose each of $N=200$ consecutive sections has uniform thickness $\Delta z=50\mathrm{nm}$, with lateral pixel pitch $p=4\mathrm{nm}$. Ignoring lost material and thickness variation, the sections coverThe voxel sampling is 12.5 times coarser along the cutting direction. These are chosen dimensions, not measurements from either cited experiment. The first and last section centers are only $(N-1)\Delta z$ apart: center-to-center extent differs from total represented thickness. Neither spacing is automatically the achieved spatial resolution.

Image resampling onto a cubic grid can make an image stack easier to display, but interpolation supplies no additional independent observations. Likewise, smooth image registration cannot resolve missing-section ambiguity in volume reconstruction: a small connecting feature could lie wholly inside lost material. Checking a reconstructed connection therefore needs evidence in the acquired slices, not just a visually smooth rendered surface.

A historically important automation change illustrates the distinction between section and surface imaging. In 2004, Winfried Denk and Heinz Horstmann placed an ultramicrotome with a diamond knife inside a scanning electron microscope. Their serial block-face scanning electron microscopy imaged the exposed block surface between cuts instead of collecting the removed sections. They reported stacks several hundred layers deep, 50–70 nm sectioning, and typically less than 10 nm lateral position jitter. These observations tested an apparatus for acquiring closely aligned serial images; they did not remove the finite spacing between observations.^[ref]

A mechanical detail makes the measurement chain concrete. Their specimen-advance lever reduced the motorized micrometer's motion by 0.11. Thus its stated position measurement uncertainty of 50 nm contributed $0.11\times 50=5.5\mathrm{nm}$ after that reduction; other sources of section-thickness error still required assessment.^[ref]

Ultramicrotomy prepares extremely thin physical sections by advancing a specimen block through a controlled increment and cutting across it with an ultramicrotome. A diamond knife or suitable glass knife removes the section; collected sections can then be examined by transmission electron microscopy. The instrument's feed setting and the geometry of the recovered section are related quantities, but they need not be equal. JEOL explicitly notes dependence on specimen material and temperature.^[ref]

For a materials example, consider a polymer block containing an initially $20.0\mu \mathrm{m}$-long marked interval parallel to the cutting direction. Suppose the recovered section measures $17.0\mu \mathrm{m}$ along that interval, with unchanged transverse width. Its section compression is $1-17/20=15\%$. Under an explicitly ideal volume-conserving model, a nominal $60\mathrm{nm}$ layer would then have recovered thicknessThe altered area and thickness give the same model volume: $0.85\cdot 70.6\simeq 60$. This is an original diagnostic calculation, not a universal correction rule. Material loss, transverse strain or nonuniform deformation would invalidate the model.

A motivating preparation check therefore compares dimensions of recognizable material features before cutting and in the recovered section, while recording cutting direction and specimen conditions. An independent thickness measurement tests whether the feed is a good depth calibration. Image registration can align section images but cannot generally reverse unknown physical deformation. For serial-section electron microscopy, preserve section order and gaps in the image stack as well as the chosen image stack axis calibration.