Merge commit '73f40ecaf0b24c44cc3accf0eb22f367b7f84070' into violet

docs/Javascript_video_analysis.md

@@ -52,6 +52,7 @@ Ranked in descending order of viability:
 
 ## Machine vision libraries
 
+## Machine vision libraries
 Top pics, maybe useful
 
 * [tracking.js](http://trackingjs.com/) does cool stuff already
@@ -67,7 +68,6 @@ Suspected non-useful or abandoned
 
 ## WebGL optimisation
 
-* notes about webgl support - e.g.
 * FFT options for mobile.
 
     * [webgl fft paper](http://www.wuhao.co/uploads/2/6/0/1/26012804/paper_final.pdf)

docs/SUMMARY.md

@@ -6,10 +6,10 @@
 * [Video analysis theory](Video_analysis.md)
 * [How to to develop synestizer](HOWTO_develop.md)
   * [Javascript libraries](Javascript_details.md)
-    * [Javascript SVG](SVG_HOWTO.md)
-   * [Video analysis in javascript](javascript_video_analysis.md)
-   * [Audio synthesis](Audio_synthesis.md)
-   * [TODO list](TODO.md)
-   * [Miscellaneous notes](Misc_notes.md)
+  * [Javascript SVG](SVG_HOWTO.md)
+  * [Video analysis in javascript](javascript_video_analysis.md)
+  * [Audio synthesis](Audio_synthesis.md)
+  * [TODO list](TODO.md)
+  * [Miscellaneous notes](Misc_notes.md)
 * [FAQ](FAQ.md)
 * [License](LICENSE.md)

docs/Video_analysis.md

@@ -97,15 +97,12 @@ In Synestizer, these $$ s_i $$ parameters are represented as the slider position
     * would the color coordinates interact at all?
     * could even takes squared difference features to extract localised frequency
 
-* something like a convolution layer from machine vision?
 * autocorrelation
 * particle filters
 * FFT features (or something else translation/phase-invariant?)
 * inner-product with desired eigen-features
 * user interaction: They choose a few key scenes, and we try to measure distance from those scenes.
 * other clustering, say, spectral?
-* [motion detection](http://www.adobe.com/devnet/html5/articles/javascript-motion-detection.html)
-* [various sweet segmenations hacks](https://stackoverflow.com/questions/31071781/html5-canvas-image-segmentation)
 * neural networks?
 
     * We can [train them online](https://cs.stanford.edu/people/karpathy/convnetjs/)
@@ -122,37 +119,6 @@ In Synestizer, these $$ s_i $$ parameters are represented as the slider position
     * http://techtalks.tv/talks/randomized-decision-forests-and-their-applications-in-computer-vision-jamie/59432/
 
 
-## Machine vision libraries
-
-* tangible.js tracks a bunch of useful libraries such as [machine vision in js](http://tangiblejs.com/libraries/computer-vision)
-* [tracking.js](http://trackingjs.com/) does cool stuff already
-* so does [jsfeat](https://inspirit.github.io/jsfeat/) including some by EPFL and an entire linear algebra library
-* [js-objectdetect](https://github.com/mtschirs/js-objectdetect/) also looks decent and fast
-* [opencvjs](https://github.com/blittle/opencvjs) looks abandoned
-* [blob detection](http://blog.acipo.com/blob-detection-js/)
-* [segmentation engine](http://vision.akshaybhat.com/)
-* [graphcut](http://www.jscuts.com/graphcuts/)
-
-## WebGL optimisation
-
-* notes about webgl support - e.g.
-* FFT options for mobile.
-
-    * [webgl fft paper](http://www.wuhao.co/uploads/2/6/0/1/26012804/paper_final.pdf)
-    * [webgl fft demo](https://github.com/wuhao1117/WebGL-Ocean-FFT). No open-source, sadly.
-    * [MDC animates textures in webgl ](https://developer.mozilla.org/en-US/docs/Web/WebGL/Animating_textures_in_WebGL)
-
-* video+WebGL:
-
-    * [live video in webgl](http://learningthreejs.com/blog/2012/02/07/live-video-in-webgl/)
-    * [three.js and video](http://threejs.org/examples/#canvas_materials_video)
-
-* [how to do WebGL-optimized image processing](http://learningwebgl.com/blog/?p=1786)
-* [good plain intro](http://www.html5rocks.com/en/tutorials/webgl/webgl_fundamentals/)
-* [webgl transforms](http://games.greggman.com/game/webgl-2d-matrices/)
-
-* CSS filters to shunt to GPU? blur+invert+opacity gives us a cheap edge detection
-
 ## Colour handling
 
 * [colorspaces](https://vis4.net/blog/posts/avoid-equidistant-hsv-colors/)