Related
I'm working on an art project which converts pixels of live video feed into corporate logos based on the distance (in RGB) between the colors of the two. While this functions, it gives a jittery result. Seem like some points in the color space teeter in a sort of superposition between two "closest" points. I'm attempting a sort of naive clustering solution right now because I believe any proper one will be too slow for live video. I'm wondering if anyone has any good ideas to solve this problem? I'll include my code and an example of the result. Thank you!
(imgs array is the logos)
current result: https://gifyu.com/image/fk2y
function distance(r1, g1, b1, bright1, r2, g2, b2, bright2) {
d =
((r2 - r1) * 0.3) ** 2 +
((g2 - g1) * 0.59) ** 2 +
((b2 - b1) * 0.11) ** 2 +
((bright2 - bright1) * 0.75) ** 2;
return Math.round(d);
}
function draw() {
if (x > 100 && z == true) {
video.loadPixels();
for (var y = 0; y < video.height; y++) {
for (var x = 0; x < video.width; x++) {
var index = (video.width - x - 1 + y * video.width) * 4;
var r = video.pixels[index];
var g = video.pixels[index + 1];
var b = video.pixels[index + 2];
var bright = (r + g + b) / 3;
let least = 9999999;
for (var i = 0; i < imgs.length; i++) {
if (
distance(
imgs[i].r,
imgs[i].g,
imgs[i].b,
imgs[i].bright,
r,
g,
b,
bright
) < least
) {
least = distance(
imgs[i].r,
imgs[i].g,
imgs[i].b,
imgs[i].bright,
r,
g,
b,
bright
);
place = imgs[i].img;
}
}
image(place, round(x * vScale), y * vScale, vScale, vScale);
}
}
}
}
I have co-ordinates for the points by taking which I draw a polygon. I can add points dynamically on the edges of the polygon and when I drag any point it should drag only the connected lines. As points can be added later on the edges so the point co-ordinates need to be ordered/sorted and the polygon should be redrawn by taking the ordered/sorted points so that on dragging any point the lines connected to the dragged point only should be dragged/updated. So to order/sort the points I am sorting the co-ordinates(2D-points) clockwise using Graham Scan/ sorting by polar angle.
My sorting code is
I find the center of the polygon like
function findCenter(points) {
let x = 0,
y = 0,
i,
len = points.length;
for (i = 0; i < len; i++) {
x += Number(points[i][0]);
y += Number(points[i][1]);
}
return { x: x / len, y: y / len }; // return average position
}
Then I sort the points by finding angles of each point from the center like
function findAngle(points) {
const center = findCenter(points);
// find angle
points.forEach((point) => {
point.angle = Math.atan2(point[1] - center.y, point[0] - center.x);
});
}
//arrVertexes is the array of points
arrVertexes.sort(function (a, b) {
return a.angle >= b.angle ? 1 : -1;
});
But the problem I am facing is if I drag any point more towards opposite side and add a new point on the edges afterward and drag the newly added point the sorting of co-ordinates is not ordered exactly because of which there is a flickering while dragging.
Here is a pictorial view of the problem I am facing for quick understanding.
Initially my svg looks like
After this I add a point and dragged like
Then I added one more point like
once I drag the added point towards down, it redraws the polygon something like (is not it weird ?)
Actually It should be like
NOTE: I really don't know what logic should I apply to get the desire functionality. Seeking help from the community leads.
Demo App
So I am looking for a solution that won't give me weird redrawing of the lines. Only the connected lines to the dragged point should be dragged.
EDIT
I came up with MUCH BETTER solution. The only problem with this approach is, When I try to add a new point on left-vertical line and If I try to move it, that newly added point moves to top-horizontal line
Updated-Demo
I've fixed this bug with left line. Take a look: codepen.
I changed getClosestPointOnLines function (actually refactored a little):
as I understood, the result here is to get i - the index for the new point in array, so I moved the algorithm to new function getI
I changed getI to use not only n (current index), but just 2 any indexes: n1 and n2: const getI = (n1, n2) => {
So all your aXys[n] is now a1 and aXys[n - 1] is now a2.
the result of getI is return i; - this is what we want from this function
I added new function-helper updateI. It calls getI and check if there any positive result.
const updateI = (n1, n2) => {
const newI = getI(n1, n2);
if (newI !== undefined) {
i = newI;
return true;
}
};
So your loop over points is now:
for (let n = 1; n < aXys.length; n++) {
updateI(n, n - 1);
}
But we need to check "left" line separately (because it connects begin and end of the array):
if (updateI(aXys.length - 1, 0)) i = aXys.length;
Sorry, but I disabled part of your code. I did not check where do you use it:
if (i < aXys.length) {
let dx = aXys[i - 1][0] - aXys[i][0];
let dy = aXys[i - 1][1] - aXys[i][1];
x = aXys[i - 1][0] - dx * fTo;
y = aXys[i - 1][1] - dy * fTo;
}
So the final version of getClosestPointOnLines now looks like this:
function getClosestPointOnLines(pXy, aXys) {
var minDist;
var fTo;
var fFrom;
var x;
var y;
var i;
var dist;
if (aXys.length > 1) {
const getI = (n1, n2) => {
let i;
const a1 = aXys[n1];
const a2 = aXys[n2];
if (a1[0] != a2[0]) {
let a = (a1[1] - a2[1]) / (a1[0] - a2[0]);
let b = a1[1] - a * a1[0];
dist = Math.abs(a * pXy[0] + b - pXy[1]) / Math.sqrt(a * a + 1);
} else dist = Math.abs(pXy[0] - a1[0]);
// length^2 of line segment
let rl2 = Math.pow(a1[1] - a2[1], 2) + Math.pow(a1[0] - a2[0], 2);
// distance^2 of pt to end line segment
let ln2 = Math.pow(a1[1] - pXy[1], 2) + Math.pow(a1[0] - pXy[0], 2);
// distance^2 of pt to begin line segment
let lnm12 = Math.pow(a2[1] - pXy[1], 2) + Math.pow(a2[0] - pXy[0], 2);
// minimum distance^2 of pt to infinite line
let dist2 = Math.pow(dist, 2);
// calculated length^2 of line segment
let calcrl2 = ln2 - dist2 + lnm12 - dist2;
// redefine minimum distance to line segment (not infinite line) if necessary
if (calcrl2 > rl2) dist = Math.sqrt(Math.min(ln2, lnm12));
if (minDist == null || minDist > dist) {
if (calcrl2 > rl2) {
if (lnm12 < ln2) {
fTo = 0; //nearer to previous point
fFrom = 1;
} else {
fFrom = 0; //nearer to current point
fTo = 1;
}
} else {
// perpendicular from point intersects line segment
fTo = Math.sqrt(lnm12 - dist2) / Math.sqrt(rl2);
fFrom = Math.sqrt(ln2 - dist2) / Math.sqrt(rl2);
}
minDist = dist;
i = n1;
}
return i;
};
const updateI = (n1, n2) => {
const newI = getI(n1, n2);
if (newI !== undefined) {
i = newI;
return true;
}
};
for (let n = 1; n < aXys.length; n++) {
updateI(n, n - 1);
}
if (updateI(aXys.length - 1, 0)) i = aXys.length;
if (i < aXys.length) {
let dx = aXys[i - 1][0] - aXys[i][0];
let dy = aXys[i - 1][1] - aXys[i][1];
x = aXys[i - 1][0] - dx * fTo;
y = aXys[i - 1][1] - dy * fTo;
}
}
console.log(aXys[i - 1]);
return { x: x, y: y, i: i, fTo: fTo, fFrom: fFrom };
}
Working example on codepen.
You should not allow any point to be added that is not close to a line.
When the user clicks, use the distance from a point to a line algorithm to check each line to see if the click is within an acceptable distance of the line. Perhaps a few pixels. If more than one line is within an acceptable distance, perhaps choose the one that is closest.
You now know where in the array to insert the new point. It will be between the first and second points of the line that just matched.
If you do that, the shape drawing should just work.
I'm receiving all distances between a random number of points in a 2 dimensional coordinate system.
How can I visualize this as coordinates on a map in my browser?
In case there are many solutions I just want to see the first possible one that my algorithm can come up with.
So here's an extremely easy example:
PointCount = 3
Distances:
0-1 = 2
0-2 = 4
1-2 = 2
Does anyone know an easy way (existing solution/framework maybe) to do it using whatever is out there to make it easier to implement?
I was thinking maybe using the html canvas element for drawing, but I don't know how to create an algorithm that could come up with possible coordinates for those points.
The above example is simplified -
Real distance values could look like this:
(0) (1) (2) (3)
(0) 0 2344 3333 10000
(1) 0 3566 10333
(2) 0 12520
I'm not sure this is relevant for SO, but anyway...
The way to do this is quite simply to place the points one by one using the data:
Pick a random location for the first point (let's say it's 0,0).
The second point is on a circle with radius d(0,1) with the first point as its center, so you can pick any point on the circle. Let's pick (d(0,1),0).
The third point is at the intersection of a circle with radius d(0,2) and center point 1, and a circle with radius d(1,2) and center point 2. You will get either 0, 1, 2 or an infinity of solutions. If the data comes from real points, 0 shouldn't happen. 1 and infinity are edge cases, but you should still handle them. Pick any of the solutions.
The fourth point is at the intersection of 3 circles. Unless you're very unlucky (but you should account for it), there should be only one solution.
Continue like this until all points have been placed.
Note that this doesn't mean you'll get the exact locations of the original points: you can have any combination of a translation (the choice of your first point), rotation (the choice of your second point) and symmetry (the choice of your third point) making the difference.
A quick and dirty implementation (not handling quite a few cases, and tested very little):
function distance(p1, p2) {
return Math.sqrt(Math.pow(p2[0] - p1[0], 2) + Math.pow(p2[1] - p1[1], 2));
}
// adapted from https://stackoverflow.com/a/12221389/3527940
function intersection(x0, y0, r0, x1, y1, r1) {
var a, dx, dy, d, h, rx, ry;
var x2, y2;
/* dx and dy are the vertical and horizontal distances between
* the circle centers.
*/
dx = x1 - x0;
dy = y1 - y0;
/* Determine the straight-line distance between the centers. */
d = Math.sqrt((dy * dy) + (dx * dx));
/* Check for solvability. */
if (d > (r0 + r1)) {
/* no solution. circles do not intersect. */
return false;
}
if (d < Math.abs(r0 - r1)) {
/* no solution. one circle is contained in the other */
return false;
}
/* 'point 2' is the point where the line through the circle
* intersection points crosses the line between the circle
* centers.
*/
/* Determine the distance from point 0 to point 2. */
a = ((r0 * r0) - (r1 * r1) + (d * d)) / (2.0 * d);
/* Determine the coordinates of point 2. */
x2 = x0 + (dx * a / d);
y2 = y0 + (dy * a / d);
/* Determine the distance from point 2 to either of the
* intersection points.
*/
h = Math.sqrt((r0 * r0) - (a * a));
/* Now determine the offsets of the intersection points from
* point 2.
*/
rx = -dy * (h / d);
ry = dx * (h / d);
/* Determine the absolute intersection points. */
var xi = x2 + rx;
var xi_prime = x2 - rx;
var yi = y2 + ry;
var yi_prime = y2 - ry;
return [
[xi, yi],
[xi_prime, yi_prime]
];
}
function generateData(nbPoints) {
var i, j, k;
var originalPoints = [];
for (i = 0; i < nbPoints; i++) {
originalPoints.push([Math.random() * 20000 - 10000, Math.random() * 20000 - 10000]);
}
var data = [];
var distances;
for (i = 0; i < nbPoints; i++) {
distances = [];
for (j = 0; j < i; j++) {
distances.push(distance(originalPoints[i], originalPoints[j]));
}
data.push(distances);
}
//console.log("original points", originalPoints);
//console.log("distance data", data);
return data;
}
function findPointsForDistances(data, threshold) {
var points = [];
var solutions;
var solutions1, solutions2;
var point;
var i, j, k;
if (!threshold)
threshold = 0.01;
// First point, arbitrarily set at 0,0
points.push([0, 0]);
// Second point, arbitrarily set at d(0,1),0
points.push([data[1][0], 0]);
// Third point, intersection of two circles, pick any solution
solutions = intersection(
points[0][0], points[0][1], data[2][0],
points[1][0], points[1][1], data[2][1]);
//console.log("possible solutions for point 3", solutions);
points.push(solutions[0]);
//console.log("solution for points 1, 2 and 3", points);
found = true;
// Subsequent points, intersections of n-1 circles, use first two to find 2 solutions,
// the 3rd to pick one of the two
// then use others to check it's valid
for (i = 3; i < data.length; i++) {
// distances to points 1 and 2 give two circles and two possible solutions
solutions = intersection(
points[0][0], points[0][1], data[i][0],
points[1][0], points[1][1], data[i][1]);
//console.log("possible solutions for point " + (i + 1), solutions);
// try to find which solution is compatible with distance to point 3
found = false;
for (j = 0; j < 2; j++) {
if (Math.abs(distance(solutions[j], points[2]) - data[i][2]) <= threshold) {
point = solutions[j];
found = true;
break;
}
}
if (!found) {
console.log("could not find solution for point " + (i + 1));
console.log("distance data", data);
console.log("solution for points 1, 2 and 3", points);
console.log("possible solutions for point " + (i + 1), solutions);
console.log("distances to point 3",
distance(solutions[0], points[2]),
distance(solutions[1], points[2]),
data[i][2]
);
break;
}
// We have found a solution, we need to check it's valid
for (j = 3; j < i; j++) {
if (Math.abs(distance(point, points[j]) - data[i][j]) > threshold) {
console.log("Could not verify solution", point, "for point " + (i + 1) + " against distance to point " + (j + 1));
found = false;
break;
}
}
if (!found) {
console.log("stopping");
break;
}
points.push(point);
}
if (found) {
//console.log("complete solution", points);
return points;
}
}
console.log(findPointsForDistances([
[],
[2344],
[3333, 3566],
[10000, 10333, 12520],
]));
console.log(findPointsForDistances([
[],
[2],
[4, 2],
]));
console.log(findPointsForDistances([
[],
[4000],
[5000, 3000],
[3000, 5000, 4000]
]));
console.log(findPointsForDistances([
[],
[2928],
[4938, 3437],
[10557, 10726, 13535]
]));
var nbPoints, i;
for (nbPoints = 4; nbPoints < 8; nbPoints++) {
for (i = 0; i < 10; i++) {
console.log(findPointsForDistances(generateData(nbPoints)));
}
}
Fiddle here: https://jsfiddle.net/jacquesc/82aqmpnb/15/
Minimum working example. Remember that in canvas coordinates, the y value is inverted but you could do something like:
y = canvasHeight - y
If you also have negative points then if would take a little bit of extra work. Also it may be helpful in that case to draw lines and tick marks to visualize the axis.
let canvas = document.getElementById("canvas");
let ctx = canvas.getContext("2d");
let scale = 10;
let radius = 10;
function point(x, y) {
ctx.fillRect(x*scale, y*scale, radius, radius);
}
// test
point(10, 15);
point(20, 8);
<html>
<body>
<canvas id="canvas" width=1000 height=1000></canvas>
</body>
</html>
There are plenty of libraries out there.
chartist.js is easy to use and responsive JavaS cript library. I used it last year for basic charts after trying many others but it was the only one that scaling easily in different screen sizes.
chartJS is another better looking library.
And you can use html5 canvas it's easy and fun but it will take time especially in scaling.
To scale and position, you should use the minimum and maximum values for x and y.
Good luck
I have a Dataset (~ 100mb) and want to get a better understanding of the data by first visualizing the amount of different JSON values.
I started by drawing an arc with ctx.arc(); and increasing the radius for each occurrence of a value.
switch(data[i].value) {
case "X":
ctx.beginPath();
ctx.arc(x, 100, i+1, 0, 2*Math.PI);
ctx.fillStyle = "MidnightBlue";
ctx.fill();
break;
}
The arc is drawn but way to big and goes beyond my viewport. So it seems that a) I'm making a mistake or b) there are just to much occurrences of the value, which cause the circle to become gigantic. How could I counter that problem?
Visualizing large values
There are two ways to visualize data that has large values.
You have given no clue as to the structure of the data so mostly I am just guessing about the data.
Scale and translate.
If the distribution of values is roughly linear you can scale and move the values to fit within the needed range.
To do this you go thought all the data points one at a time and find the minimum value and max value.
var min = Infinity; // set the start min and max
var max = -Infinity;
for (var i = 0; i < data.length; i++){
if (data[i].value === "X") {
// I dont know where you got the x from so leave that to you
// x is the value needed to graph
min = Math.min(min, x);
max = Math.max(max, x);
}
}
After you have check each value and have a min and max you need to workout how big you want to display the info.
const displayMaxRadius = Math.min(canvas.width, canvas.height) / 2;
const displayMinRadius = 10;
Then to display each value you use the min and max to scale to a normalized range, bringing each value to be within 0 to 1 inclusive. The scale to fit te display min and max
for (var i = 0; i < data.length; i ++) {
if (data[i].value === "X") {
// I dont know where you got the x from so leave that to you
// x is the value needed to graph
var norm = (x - min) / (max - min); // normalize the value
var displaySize = norm * (displayMaxRadius - displayMinRadius) + displayMinRadius;
ctx.beginPath();
ctx.arc(displaySize , 100, i + 1, 0, 2 * Math.PI);
ctx.fillStyle = "MidnightBlue";
ctx.fill();
Logarithmic data
Sometimes the range of values is spread unevenly over a very large range, with clumps of data at some ranges. Using the above method will work but for most of the data it will be scaled such that the individual differences are lost due to the large range of values.
To deal with that you create a logarithmic graph, simple find the root of the values before you find the min max range. You can use the square root or to any other value.
Use Math.pow(x,1/r) where r is to what root you want r = 2 is square root, r = 3 is cubic root, and so on
var root = 2; // sqrt root
var min = Infinity; // set the start min and max
var max = -Infinity;
for (var i = 0; i < data.length; i++) {
if (data[i].value === "X") {
// I dont know where you got the x from so leave that to you
// x is the value needed to graph
var rval = Math.pow(x, root);
min = Math.min(min, rval);
max = Math.max(max, rval);
}
}
for (var i = 0; i < data.length; i++) {
if (data[i].value === "X") {
// I dont know where you got the x from so leave that to you
// x is the value needed to graph
var rval = Math.pow(x, root);
var norm = (rval - min) / (max - min); // normalize the value
var displaySize = norm * (displayMaxRadius - displayMinRadius) + displayMinRadius;
ctx.beginPath();
ctx.arc(displaySize , 100, i + 1, 0, 2*Math.PI);
ctx.fillStyle = "MidnightBlue";
ctx.fill();
I sort of found an answer to my question myself. What I could do is create a grid/rectangle with modulo.
var x = (i % 115) * 1;
var y = Math.floor(i / 115) * 1;
ctx.fillStyle = "MidnightBlue";
ctx.fillRect(x, y, 1, 1);
As you can see I have key/value pairs of the states in the US. To visualize the occurrence of each state in the dataset I want to draw a grid with modulo.
The number 115 is the root of 13450. But 13450 is the amount of (for example) all farms in the US. Now I want to visualize just the farms in PA … How could I do that?
I've been working on this problem for some time now with little promising results. I am trying to split up an image into connected regions of similar color. (basically split a list of all the pixels into multiple groups (each group containing the coordinates of the pixels that belong to it and share a similar color).
For example:
http://unsplash.com/photos/SoC1ex6sI4w/
In this image the dark clouds at the top would probably fall into one group. Some of the grey rock on the mountain in another, and some of the orange grass in another. The snow would be another - the red of the backpack - etc.
I'm trying to design an algorithm that will be both accurate and efficient (it needs to run in a matter of ms on midrange laptop grade hardware)
Below is what I have tried:
Using a connected component based algorithm to go through every pixel from top left scanning every line of pixels from left to right (and comparing the current pixel to the top pixel and left pixel). Using the CIEDE2000 color difference formula if the pixel at the top or left was within a certain range then it would be considered "similar" and part of the group.
This sort of worked - but the problem is it relies on color regions having sharp edges - if any color groups are connected by a soft gradient it will travel down that gradient and continue to "join" the pixels as the difference between the individual pixels being compared is small enough to be considered "similar".
To try to fix this I chose to set every visited pixel's color to the color of most "similar" adjacent pixel (either top or left). If there are no similar pixels than it retains it's original color. This somewhat fixes the issue of more blurred boundaries or soft edges because the first color of a new group will be "carried" along as the algorithm progresses and eventually the difference between that color and the current compared color will exceed the "similarity" threashold and no longer be part of that group.
Hopefully this is making sense. The problem is neither of these options are really working. On the image above what is returned are not clean groups but noisy fragmented groups that is not what I am looking for.
I'm not looking for code specifically - but more ideas as to how an algorithm could be structured to successfully combat this problem. Does anyone have ideas about this?
Thanks!
You could convert from RGB to HSL to make it easier to calculate the distance between the colors. I'm setting the color difference tolerance in the line:
if (color_distance(original_pixels[i], group_headers[j]) < 0.3) {...}
If you change 0.3, you can get different results.
See it working.
Please, let me know if it helps.
function hsl_to_rgb(h, s, l) {
// from http://stackoverflow.com/questions/2353211/hsl-to-rgb-color-conversion
var r, g, b;
if (s == 0) {
r = g = b = l; // achromatic
} else {
var hue2rgb = function hue2rgb(p, q, t) {
if (t < 0) t += 1;
if (t > 1) t -= 1;
if (t < 1 / 6) return p + (q - p) * 6 * t;
if (t < 1 / 2) return q;
if (t < 2 / 3) return p + (q - p) * (2 / 3 - t) * 6;
return p;
}
var q = l < 0.5 ? l * (1 + s) : l + s - l * s;
var p = 2 * l - q;
r = hue2rgb(p, q, h + 1 / 3);
g = hue2rgb(p, q, h);
b = hue2rgb(p, q, h - 1 / 3);
}
return [Math.round(r * 255), Math.round(g * 255), Math.round(b * 255)];
}
function rgb_to_hsl(r, g, b) {
// from http://stackoverflow.com/questions/2353211/hsl-to-rgb-color-conversion
r /= 255, g /= 255, b /= 255;
var max = Math.max(r, g, b),
min = Math.min(r, g, b);
var h, s, l = (max + min) / 2;
if (max == min) {
h = s = 0; // achromatic
} else {
var d = max - min;
s = l > 0.5 ? d / (2 - max - min) : d / (max + min);
switch (max) {
case r:
h = (g - b) / d + (g < b ? 6 : 0);
break;
case g:
h = (b - r) / d + 2;
break;
case b:
h = (r - g) / d + 4;
break;
}
h /= 6;
}
return [h, s, l];
}
function color_distance(v1, v2) {
// from http://stackoverflow.com/a/13587077/1204332
var i,
d = 0;
for (i = 0; i < v1.length; i++) {
d += (v1[i] - v2[i]) * (v1[i] - v2[i]);
}
return Math.sqrt(d);
};
function round_to_groups(group_nr, x) {
var divisor = 255 / group_nr;
return Math.ceil(x / divisor) * divisor;
};
function pixel_data_to_key(pixel_data) {
return pixel_data[0].toString() + '-' + pixel_data[1].toString() + '-' + pixel_data[2].toString();
}
function posterize(context, image_data, palette) {
for (var i = 0; i < image_data.data.length; i += 4) {
rgb = image_data.data.slice(i, i + 3);
hsl = rgb_to_hsl(rgb[0], rgb[1], rgb[2]);
key = pixel_data_to_key(hsl);
if (key in palette) {
new_hsl = palette[key];
new_rgb = hsl_to_rgb(new_hsl[0], new_hsl[1], new_hsl[2]);
rgb = hsl_to_rgb(hsl);
image_data.data[i] = new_rgb[0];
image_data.data[i + 1] = new_rgb[1];
image_data.data[i + 2] = new_rgb[2];
}
}
context.putImageData(image_data, 0, 0);
}
function draw(img) {
var canvas = document.getElementById('canvas');
var context = canvas.getContext('2d');
context.drawImage(img, 0, 0, canvas.width, canvas.height);
img.style.display = 'none';
var image_data = context.getImageData(0, 0, canvas.width, canvas.height);
var data = image_data.data;
context.drawImage(target_image, 0, 0, canvas.width, canvas.height);
data = context.getImageData(0, 0, canvas.width, canvas.height).data;
original_pixels = [];
for (i = 0; i < data.length; i += 4) {
rgb = data.slice(i, i + 3);
hsl = rgb_to_hsl(rgb[0], rgb[1], rgb[2]);
original_pixels.push(hsl);
}
group_headers = [];
groups = {};
for (i = 0; i < original_pixels.length; i += 1) {
if (group_headers.length == 0) {
group_headers.push(original_pixels[i]);
}
group_found = false;
for (j = 0; j < group_headers.length; j += 1) {
// if a similar color was already observed
if (color_distance(original_pixels[i], group_headers[j]) < 0.3) {
group_found = true;
if (!(pixel_data_to_key(original_pixels[i]) in groups)) {
groups[pixel_data_to_key(original_pixels[i])] = group_headers[j];
}
}
if (group_found) {
break;
}
}
if (!group_found) {
if (group_headers.indexOf(original_pixels[i]) == -1) {
group_headers.push(original_pixels[i]);
}
if (!(pixel_data_to_key(original_pixels[i]) in groups)) {
groups[pixel_data_to_key(original_pixels[i])] = original_pixels[i];
}
}
}
posterize(context, image_data, groups)
}
var target_image = new Image();
target_image.crossOrigin = "";
target_image.onload = function() {
draw(target_image)
};
target_image.src = "http://i.imgur.com/zRzdADA.jpg";
canvas {
width: 300px;
height: 200px;
}
<canvas id="canvas"></canvas>
You can use "Mean Shift Filtering" algorithm to do the same.
Here's an example.
You will have to determine function parameters heuristically.
And here's the wrapper for the same in node.js
npm Wrapper for meanshift algorithm
Hope this helps!
The process you are trying to complete is called Image Segmentation and it's a well studied area in computer vision, with hundreds of different algorithms and implementations.
The algorithm you mentioned should work for simple images, however for real world images such as the one you linked to, you will probably need a more sophisticated algorithm, maybe even one that is domain specific (are all of your images contains a view?).
I have little experience in Node.js, however from Googling a bit I found the GraphicsMagic library, which as a segment function that might do the job (haven't verified).
In any case, I would try looking for "Image segmentation" libraries, and if possible, not limit myself only to Node.js implementations, as this language is not the common practice for writing vision applications, as opposed to C++ / Java / Python.
I would try a different aproach. Check out this description of how a flood fill algorithm could work:
Create an array to hold information about already colored coordinates.
Create a work list array to hold coordinates that must be looked at. Put the start position in it.
When the work list is empty, we are done.
Remove one pair of coordinates from the work list.
If those coordinates are already in our array of colored pixels, go back to step 3.
Color the pixel at the current coordinates and add the coordinates to the array of colored pixels.
Add the coordinates of each adjacent pixel whose color is the same as the starting pixel’s original color to the work list.
Return to step 3.
The "search approach" is superior because it does not only search from left to right, but in all directions.
You might look at k-means clustering.
http://docs.opencv.org/3.0-beta/modules/core/doc/clustering.html