chickadee » image-processing


This page is maintained in the package's github repository.

This is a Chicken Scheme egg with various image processing utilities.

Note that a few of these functions call ImageMagick.

Color spaces


These are exposed in case we want to switch to 16-bit grayscale or colors.

rgb->hsv rgbprocedure
hsv->rgb hsvprocedure
rgb->cd rgbprocedure
rgb->cmyk rgbprocedure
rgb->uv-hsv cprocedure

Various color space conversions. Colors are vectors of values.

rgb->xyz cprocedure
xyz->rgb cprocedure
xyz->l*ab cprocedure
l*ab->xyz cprocedure
rgb->l*ab cprocedure
l*ab->rgb cprocedure

Convert to l*ab. Colors are vectors of values. This is space in which Euclidean distance is similar to human perceputal distance. These implement a 2 degree observer and the D65 illuminant.

rgb->html cprocedure

Not really a color space conversion, covert an rgb color to an HTML color #rrggbb.


define-structure pbm raw? bitmaprecord
define-structure pgm raw? maxval greyrecord
define-structure ppm raw? maxval red green bluerecord
pnm? mprocedure

Three image formats are supported. PBMs are binary images, PGMs are greyscale and PPMs are tri-plane (nominally RGB) images. All planes are stored as scheme matrices. Together all of these formats are called PNM. This is a standard format that many tools (such browser and ImageMagick) can read and write. PNMs can be serialized as either raw packed binaries or ascii text files. Note that we are unable to read or write raw PBMs.

image-ref i pprocedure

Index into an PNM image or into a matrix. This is like indexing into a matrix except that the arguments are reversed. A matrix-ref takes the row and then the column, image-ref takes a point which is an (x,y) coordinate.

pbm-ascii pbmprocedure

Ensure that the PBM image is in ascii. We are unable to read or write raw PBMs.

pnm-copy pnmprocedure

The resulting image will not share any data with the original.

Creating images

read-pnm pathnameprocedure

Read a PNM. We determine the kind of image (PBM, PGM, PPM see above) from the file header.

write-pnm pnm pathnameprocedure

Write a PNM. If don't provide an extension an appropriate one will be appended to your pathname.

pbm-constant width height bitprocedure
pgm-constant width height greyprocedure
ppm-constant width height red green blueprocedure

Create constant images.

pbm-left-vertical-stripe width height leftprocedure

Create a black vertical stripe on a white image.

crop-image pnm x y width heightprocedure

Crop an image and return a new image of the given size.

pnm->ppm pnmprocedure
pnm->pgm pnmprocedure

Convert any image format to a color or greyscale image

pbm->pgm pbmprocedure
pgm->ppm pgmprocedure
pbm->ppm pbmprocedure
ppm->pgm ppmprocedure

Various image conversions.

pgm->pbm pgm thresholdprocedure
ppm->pbm ppm thresholdprocedure

Convert a color or greyscale image to a binary image with the provided threshold.

Basic image information

pnm-width pnmprocedure
pnm-height pnmprocedure

Basic image information.

ppm-hue ppmprocedure
ppm-saturation ppmprocedure
ppm-value ppmprocedure

Creates a greyscale image representing one of the HSV channels from a color image.

ppm-mean image colour-transformprocedure
ppm-covariance image colour-transformprocedure
pgm-mean imageprocedure
pgm-variance imageprocedure

Compute the mean or covariance of an image. These take an optional colour transform.

Logical and morphological operations

pbm-and pbm1 pbm2procedure
pbm-or pbm1 pbm2procedure
pbm-not pbmprocedure
pbm-xor pbm1 pbm2procedure

Basic logical operations on binary images.

pgm-absolute-difference pgm1 pgm2procedure

Compute the difference between two greyscale images.

empty-pnm? pnmprocedure

Is this image empty?

pbm-skeletonize pbmprocedure
pbm-bloat pbm nprocedure

Bloat or skeletonize a binary image.

pbm-flood pbm pointprocedure

Floor fill a binary image.

PNM movies

pnm-movie-frame-pathname pathname iprocedure
pnm-movie-length pathnameprocedure
read-pnm-movie pathnameprocedure
write-pnm-movie pnm-movie pathnameprocedure

A PNM movie is a sequence of PNMs on disk.

Image operations

overlay-pbm-on-pnm pbm pnmprocedure

The white pixels in the pbm become white in the resulting image.

pnm-black-window pnm upper-left lower-rightprocedure
pnm-white-window pnm upper-left lower-rightprocedure

Draw white or black windows on an image.

pbm-ppm-and pbm ppmprocedure
pgm-and-pbm pgm pbmprocedure

Black pixels in the PBM become black pixels in the result.

pnm-shift pnm deltaprocedure

Shift an image down and rightward by the give number of pixels.

pgm-smooth pgm sigmaprocedure

Smooth a greyscale image with the given variance.

pbm-proximity-clusterer pbm thresholdprocedure

Cluster white pixels in the binary image with the given distance threshold.

pnm-rotate pnmprocedure
pnm-flip pnmprocedure

Rotate (transpose) an image or flip (mirror) an image.

ppm-burn base mask colourprocedure

Burn the ppm with the given mask and color.

Image pairs

normal-flow-magnitude pgm1 pgm2 epsilon sigma sensitivityprocedure
threshold-normal-flow-magnitude pgm1 pgm2 epsilon sigma thresholdprocedure

Compute normalized flow between two pairs of greyscale images.


flatten-ppm ppm colour-transformprocedure

Returns a single matrix from a tri-plane PPM. This maps colour-transform over the image.

binary-threshold pgm thresholdprocedure

Threshold a greyscale image.

find-threshold-otsu pgmprocedure
binary-threshold-optimal pgmprocedure

Otsu's Method to find an optional threshold and apply it.

find-threshold-means pgmprocedure
binary-threshold-means pgmprocedure

Method of Successive Means to find an optimal threshold and apply it.

Colour Threshold

sample-image ppm colour-tx window-centre window-sizeprocedure

Map colour-tx over an region of a color image.

binary-threshold-colour ppm colour-tx point thresholdprocedure

Binary threshold a colour image by sampling a region around the given point and thresholding all values that are within the given distance threshold of the mean color of that region.

colour-threshold ppm colour-tx mu isigma thresholdprocedure

Binary threshold a colour image with the given mean and variance. The mahalanobis-distance between each pixel and the given mean and variance will be computed and compared to the threshold.

Histogram Equalization

histogram-equalise pgmprocedure

Equalize a greyscale image.

Adaptive Thresholding

make-integral-matrix matrixprocedure

Given a matrix compute an integral matrix which can provide the mean value of a rectangular region in O(1).

compute-integral-matrix-mu integral-matrix x y wprocedure
compute-integral-matrix-sigma squared-integral-matrix mu x y wprocedure

This computes the mean and variance of a square of size w in the given integral matrix.

compute-adaptive-threshold integral-matrix squared-integral-matrix x y wprocedure

Compute an adaptive threshold on the given integral matrix.

adaptive-threshold pgm winsizeprocedure

Compute an adaptive threshold for a greyscale image with a given window size.

slide-window m f sizeprocedure

Map f over a matrix passing it square submatrices of the given size. At the edges f is passed #f.

count-pixels aprocedure

Count number of true values in a matrix.

ppm->label-closest ppm mu1 sigma1 mu2 sigma2 #!optional colour-transformprocedure

Create a binary image from a color image where points are either white or black depending on which is closer to the given means and variances.


find-histogram pixmap maxvalprocedure
histogram-mean histogram iprocedure
histogram-variance histogram mu iprocedure
normalised-histogram histogram valprocedure
weighted-histogram histogramprocedure
cumulative-histogram histogramprocedure
find-between-class-variances omegas mus mu-totalprocedure

Compute histograms from a matrix.

Rendering Line Segments

midpoint lprocedure

Compute the midpoint of a line segment.

orientation vprocedure

Compute the orientation of a vector.

line-segment-orientation lprocedure

Compute the orientation of a line segment.

line-segment->points lprocedure
line-segments->points lsprocedure
points->line-segments psprocedure

Convert between points and line semgents.

Points and bounding boxes

points->pbm-of-size points height widthprocedure

Take the list of points and create a binay image of the given size where those points are white.

points-bounding-box pointsprocedure

Computing a bounding box for a set of points.

bounding-box-size bbprocedure

Compute the size of a bounding box

points->points-bounding-box points bbprocedure

Recompute the coordinates of the given points relative to the top-left corner of the bounding box.

points->bounding-box-points pointsprocedure

Convert points to a list of points on the bounding box of those points.

normalize-to-bounding-box psprocedure
normalize-to-other-bounding-box points psprocedure
points->target-bounding-box points target-bbprocedure
points->other-target-bounding-box points ps target-bbprocedure

Normalize the cooredinates of these points to their bounding box. Or to the bounding box of the given points.

bounding-box-bloat bb pprocedure

Bloat a bounding box.

bounding-box-crop bb imageprocedure

Crop a bounding box out of an image.

pbm->points pbmprocedure
points->pbm points height widthprocedure

Convert a binary image to a set of points or vice versa.


quantize-coordinate xprocedure
quantize-point pprocedure
quantize-points psprocedure
quantize-line-segment lprocedure
quantize-line-segments lsprocedure

Quantize coordinates, points or line segments.


define-structure ellipse x0 y0 t0 a bprocedure
ellipse-center ellipseprocedure
ellipse-area ellipseprocedure
ellipse-eccentricity ellipseprocedure

Define and compute basic properties of ellipses.

radial-distance theta phiprocedure

The radial distance between two angles.

point-on-ellipse? p ellipse toleranceprocedure

Compute if a point is on an ellipse.

line-segment->ellipse lprocedure

Place an ellipse around a line segment

ellipse->points e #!optional (n 360)procedure

Sample n points on an ellipse.


resize-image w h iprocedure

Resize an image. This is quite slow as it writes the image out and calls imagemagick to resize it.

subsample-pbm pbmprocedure

Subsample a pbm by a factor of 2.

scale-ppm ppm scaleprocedure

Scale a pbm by a factor of 2 or 4.

Connected Components

connected-component-filter pbm delta thresholdprocedure

Compute connected components of a binary image with distance thershold delta and size threshold threshold.x

define-private-structure vertex pixels vertex edgesrecord
define-private-structure edge u vrecord
define-private-structure graph vertices edgesrecord

These are defined private for now as they need to be renamed.

pbm->graph pbm deltaprocedure
labeling->graph labeling deltaprocedure

Convert a binary image to a graph where pixels within delta of each other are connected.

dereference-vertex uprocedure
connected-components gprocedure

Compute the connect components of a graph.

vertices->pbm vertices height widthprocedure
graph->pbm g height widthprocedure

Convert a graph or list of vertices to a pbm.


pbm->chains pbmprocedure

Convert a binary image to a list of chains, pixels that are connected to each other.

chains->pbm chains height widthprocedure

Convert a list of chains to a pbm.

chain-filter pbm thresholdprocedure

Filters chains below a certain length.

break-chain chain lprocedure
break-chains chains lprocedure

Break chains below a certain length.


conjure pbms delta span threshold1 threshold2procedure

Fill in missing values in pairs of frames.

Distance Transform

distance-transform pbmprocedure
closest-transform-ref closest-transform pprocedure

Compute the Manhattan distance transform for a binary image and reference into that distance transform.

euclidean-1d-dt vprocedure
euclidean-2d-dt mprocedure
euclidean-1d-dt-vals vprocedure
euclidean-2d-dt-vals mprocedure

Compute a generalized squared-euclidean distance transform on the given vector or matrix. The -vals version also returns a map of the closest points.


define-structure pnm-buffer buffer width height pixfmt storage typerecord

A mechanism to store raw image pointers.

pnm->pixfmt p p4?procedure
pixfmt->stride pixfmtprocedure
image-type->pixfmt t p4?procedure
image-type->storage-size t p4?procedure
pnm->image-type pprocedure
pnm->storage pprocedure
pixfmt->red pixfmtprocedure
pixfmt->green pixfmtprocedure
pixfmt->blue pixfmtprocedure
pnm-buffer-size p p4?procedure
pnm->pnm-buffer! p #!rest p4?procedure
pnm-fill-buffer! p b pixfmtprocedure
free-pnm-buffer! pprocedure

Misc raw image buffer operations.

imlib-image->pnm-buffer! imageprocedure
imlib-image->ppm imageprocedure
ppm->imlib-image ppmprocedure

Convert between PPMs and Imlib images. PGMs and PBMs must first be converted to PPMs because Imlib only handles RGB images.

pnm-buffer->pnm bprocedure
pnm-buffer->pbm pprocedure
pnm-buffer->pgm pprocedure
pnm-buffer->ppm pprocedure

Convert a raw image buffer to a scheme image.

image->pnm-buffer! imageprocedure

Convert any image to a raw image buffer.


ppm-absolute-difference ppm1 ppm2procedure

Compute the absolute difference between two color images.

set-ppm-pixel! ppm x y valueprocedure

Set a pixel in a color image to a given value (a 3-vector).

pnm-pixel? i x yprocedure

Is this a valid position in the given image?

map-ppm-values ppm fprocedure

Map f on each component of each pixel in the image.


pbm-stack-vertical pbm1 pbm2procedure
pbm-stack-horizontal pbm1 pbm2procedure
ppm-stack-vertical ppm1 ppm2procedure

Stack pbms.

ppm-stack-horizontal ppm1 ppm2procedure

Stack ppms.


degrees->radians angleprocedure
radians->degrees angleprocedure
normalize-rotation rotationprocedure
rotation+ x yprocedure
rotation- x yprocedure
angle-separation x yprocedure
mean-angle anglesprocedure

Misc operation on angles.

rotation-matrix-2d thetaprocedure

Produce the 2D rotation matrix for the given angle.


show iprocedure

Display an image. This invokes feh.


  Written by the
  [[Center for the Computational Study of Cognition, Control, and Perception|]].
  Maintainer: Andrei Barbu,
  Copyright 1993-1995 University of Toronto. All rights reserved.
  Copyright 1996 Technion. All rights reserved.
  Copyright 1996 and 1997 University of Vermont. All rights reserved.
  Copyright 1997-2001 NEC Research Institute, Inc. All rights reserved.
  Copyright 2002-2013 Purdue University. All rights reserved.
  This program is free software: you can redistribute it and/or modify
  it under the terms of the GNU Lesser General Public License as published by
  the Free Software Foundation, either version 3 of the License, or
  (at your option) any later version.
  This program is distributed in the hope that it will be useful,
  but WITHOUT ANY WARRANTY; without even the implied warranty of
  GNU Lesser General Public License for more details.
  You should have received a copy of the GNU Lesser General Public License
  along with this program.  If not, see

Contents »