diff --git a/chapters/displays-01-principles.qmd b/chapters/displays-01-principles.qmd index 810eb2a..2bf1d24 100644 --- a/chapters/displays-01-principles.qmd +++ b/chapters/displays-01-principles.qmd @@ -5,26 +5,17 @@ date: last-modified # Display principles {#sec-display-principles} {{< include "includes/WIP-callout.qmd" >}} - -## Display principles overview {#sec-display-principles-overview} +[^ip-snark]: Computer scientists generally think RGB images magically appear and reflect reality. Sigh. IMHO there could be a lot more thought put into the question of image processing for machine vision applications—for example, standard RGB pipelines bake in human-centric demosaicking, white balance, and tone curves that are often unnecessary or suboptimal for machine vision. +## Displays - A Review chapter **The text below is adapted from this review chapter. At this time, please also consult that article:** **[Characterization of Visual Stimuli using the Standard Display Model, 2024](resources/Artal-2024/Artal-2024.qmd){target=_blank}** diff --git a/chapters/imgprocessing-01-principles.qmd b/chapters/imgprocessing-01-principles.qmd index 829b484..bc172e1 100644 --- a/chapters/imgprocessing-01-principles.qmd +++ b/chapters/imgprocessing-01-principles.qmd @@ -7,30 +7,32 @@ date: last-modified ## Image processing principles overview {#sec-imgprocessing-principles-overview} -The general field of digital image processing is enormous. It was large when I was a graduate student, and with time it has grown even more massive. It seems that whenever a new computational concept emerges, it finds application in imaging and expands the field. The most recent expansion is the use of deep neural network tools to create and edit images and videos. These are fascinating and +This chapter bridges the raw, mosaicked sensor measurements introduced in the sensor chapters and the trichromatic display representations introduced in @sec-display-principles-overview. -In +![Image processing stages](images/imageprocessing/01-image-processing/004-reference-consumer-photography-pipeline.png){#fig-ip-overview width="70%"} -To this point, we have considered the environmental light field [@sec-light-overview], how it is measured by the optics [@sec-geometric-optics], and how it is then converted into electrons at each pixel [@sec-pixels-overview]. We have also considered how the visual system would capture the same light field. +## Lens shading -The first part of this chapter on image processing describes the steps that converts the data measured by the sensor into a signal that can be viewed by a person. +Make a panel tabset of these. -It seems as if we should have display go first so we can do it. Or we might just assume that people know displays are typically RGB and hope for the best? +![Lens shading geometry.](images/imageprocessing/01-image-processing/013-relative-illumination-aperture-geometry.png){#fig-lens-shading} -Also, the human section on color might include some minimal display description that we could use here as the basic display model, with more advanced displays to come later, in the chapter @sec-display-principles-overview. -## Lens shading +![Lens shading geometry.](images/imageprocessing/01-image-processing/014-relative-illumination-sensor-distance.png){#fig-lens-shading} ## Acquisition policies + ## Demosaicking Space-color correlations? ## Color processing + ### Sensor correction + ### Illuminant correction @@ -39,3 +41,4 @@ Space-color correlations? + diff --git a/chapters/part-processing.qmd b/chapters/part-processing.qmd index d12504a..bc8caf4 100644 --- a/chapters/part-processing.qmd +++ b/chapters/part-processing.qmd @@ -7,7 +7,7 @@ date: last-modified Intro to processing. -## Image processing topics +## Image processing topics {#sec-part-ip-overview} * Hardware - Qualcomm, Broadcom type devices for rendering * Rendering - Algorithms such as demosaic, defective pixel, color transforms, blurring, adaptive