如何解决使用canvas html5为图像绘制线条轮廓
我正在尝试为图像绘制强度分布图,其中x轴为图像上线条的长度,y轴为强度值,沿着线条的长度。如何在html 5 canvas上执行此操作?我尝试了以下代码,但没有获得正确的强度值。不知道我要去哪里错了。
private getLineIntensityVals = function (lineObj,img) {
const slope = this.calculateSlopeOfLine(lineObj.upPos,lineObj.downPos);
const intercept = this.calculateIntercept(lineObj.downPos,slope);
const ctx = img.getContext('2d');
const coordinates = [];
const intensities = [];
for (let x = lineObj.downPos.x; x <= lineObj.upPos.x; x++) {
const y = slope * x + intercept;
const pixelData = ctx.getImageData(x,y,1,1).data;
pixelData[0] = 255 - pixelData[0];
pixelData[1] = 255 - pixelData[1];
pixelData[2] = 255 - pixelData[2];
const intensity = ((0.299 * pixelData[0]) + (0.587 * pixelData[1]) + (0.114 * pixelData[2]));
intensities.push(intensity);
}
return intensities;
};
private calculateSlopeOfLine = function (upPos,downPos) {
if (upPos.x === downPos.x || upPos.y === downPos.y) {
return null;
}
return (downPos.y - upPos.y) / (downPos.x - upPos.x);
};
private calculateIntercept = function (startPoint,slope) {
if (slope === null) {
return startPoint.x;
}
return startPoint.y - slope * startPoint.x;
};
private calculateLineLength(line) {
const dim = {width: Math.abs(line.downPos.x -line.upPos.x),height:Math.abs(line.downPos.y- line.upPos.y)};
length = Math.sqrt(Math.pow(dim.width,2) + Math.pow(dim.height,2));
return length;
};
解决方法
图像数据
不要一次获取一个像素的图像数据。获得对像素数据的访问成本很高(CPU周期),而内存则很便宜。一次获取所有像素,然后重复使用该数据。
数据采样
大多数线条不能均匀地放入像素中。要解决的问题是将行划分为所需的样本数(可以使用行长)
然后依次转到每个样本,获取4个相邻像素值并在样本点处插值颜色。
在进行内插时,我们需要确保不使用错误的颜色模型。在这种情况下,我们使用sRGB。
我们因此得到了功能
// imgData is the pixel date
// x1,y1 and x2,y2 are the line end points
// sampleRate is number of samples per pixel
// Return array 3 values for each sample.
function getProfile(imgData,x1,y1,x2,y2,sampleRate) {
// convert line to vector
const dx = x2 - x1;
const dy = y2 - y1;
// get length and calculate number of samples for sample rate
const samples = (dx * dx + dy * dy) ** 0.5 * Math.abs(sampleRate) + 1 | 0;
// Divide line vector by samples to get x,and y step per sample
const nx = dx / samples;
const ny = dy / samples;
const w = imgData.width;
const h = imgData.height;
const pixels = imgData.data;
const values = [];
// Offset line to center of pixel
var x = x1 + 0.5;
var y = y1 + 0.5;
var i = samples;
while (i--) { // for each sample
// make sure we are in the image
if (x >= 0 && x < w - 1 && y >= 0 && y < h - 1) {
// get 4 closest pixel indexes
const idxA = ((x | 0) + (y | 0) * w) * 4;
const idxB = ((x + 1 | 0) + (y | 0) * w) * 4;
const idxC = ((x + 1 | 0) + (y + 1 | 0) * w) * 4;
const idxD = ((x | 0) + (y + 1 | 0) * w) * 4;
// Get channel data using sRGB approximation
const r1 = pixels[idxA] ** 2.2;
const r2 = pixels[idxB] ** 2.2;
const r3 = pixels[idxC] ** 2.2;
const r4 = pixels[idxD] ** 2.2;
const g1 = pixels[idxA + 1] ** 2.2;
const g2 = pixels[idxB + 1] ** 2.2;
const g3 = pixels[idxC + 1] ** 2.2;
const g4 = pixels[idxD + 1] ** 2.2;
const b1 = pixels[idxA + 2] ** 2.2;
const b2 = pixels[idxB + 2] ** 2.2;
const b3 = pixels[idxC + 2] ** 2.2;
const b4 = pixels[idxD + 2] ** 2.2;
// find value at location via linear interpolation
const xf = x % 1;
const yf = y % 1;
const rr = (r2 - r1) * xf + r1;
const gg = (g2 - g1) * xf + g1;
const bb = (b2 - b1) * xf + b1;
/// store channels as uncompressed sRGB
values.push((((r3 - r4) * xf + r4) - rr) * yf + rr);
values.push((((g3 - g4) * xf + g4) - gg) * yf + gg);
values.push((((b3 - b4) * xf + b4) - bb) * yf + bb);
} else {
// outside image
values.push(0,0);
}
// step to next sample
x += nx;
y += ny;
}
return values;
}
转换为值
该数组保存原始样本数据。有多种方法可以转换为值。这就是为什么我们将抽样从转换到值中分开。
下一个函数获取原始样本数组并将其转换为值。它返回一个值数组。在进行转换时,它还会获得最大值,以便可以绘制数据以适合图形。
function convertToMean(values) {
var i = 0,v;
const results = [];
results._max = 0;
while (i < values.length) {
results.push(v = (values[i++] * 0.299 + values[i++] * 0.587 + values[i++] * 0.114) ** (1/2.2));
results._max = Math.max(v,results._max);
}
return results;
}
现在您可以按自己的喜好绘制数据了。
示例
点击图片上的拖动线(加载时)
结果是实时绘制的。
将鼠标移到绘图上可以查看值。
使用整页查看全部内容。
const ctx = canvas.getContext("2d");
const ctx1 = canvas1.getContext("2d");
const SCALE_IMAGE = 0.5;
const PLOT_WIDTH = 500;
const PLOT_HEIGHT = 150;
canvas1.width = PLOT_WIDTH;
canvas1.height = PLOT_HEIGHT;
const line = {x1: 0,y1: 0,x2: 0,y2:0,canUse: false,haveData: false,data: undefined};
var bounds,bounds1,imgData;
// ix iy image coords,px,py plot coords
const mouse = {ix: 0,iy: 0,overImage: false,px: 0,py:0,overPlot: false,button : false,dragging: 0};
["down","up","move"].forEach(name => document.addEventListener("mouse" + name,mouseEvents));
const img = new Image;
img.crossOrigin = "Anonymous";
img.src = "https://upload.wikimedia.org/wikipedia/commons/thumb/4/47/Black_and_yellow_garden_spider%2C_Washington_DC.jpg/800px-Black_and_yellow_garden_spider%2C_Washington_DC.jpg";
img.addEventListener("load",() => {
canvas.width = img.width;
canvas.height = img.height;
ctx.drawImage(img,0);
imgData = ctx.getImageData(0,ctx.canvas.width,ctx.canvas.height);
canvas.width = img.width * SCALE_IMAGE;
canvas.height = img.height * SCALE_IMAGE;
bounds = canvas.getBoundingClientRect();
bounds1 = canvas1.getBoundingClientRect();
requestAnimationFrame(update);
},{once: true});
function getProfile(imgData,sampleRate) {
x1 *= 1 / SCALE_IMAGE;
y1 *= 1 / SCALE_IMAGE;
x2 *= 1 / SCALE_IMAGE;
y2 *= 1 / SCALE_IMAGE;
const dx = x2 - x1;
const dy = y2 - y1;
const samples = (dx * dx + dy * dy) ** 0.5 * Math.abs(sampleRate) + 1 | 0;
const nx = dx / samples;
const ny = dy / samples;
const w = imgData.width;
const h = imgData.height;
const pixels = imgData.data;
const values = [];
var x = x1 + 0.5;
var y = y1 + 0.5;
var i = samples;
while (i--) {
if (x >= 0 && x < w - 1 && y >= 0 && y < h - 1) {
// get 4 closest pixel indexs
const idxA = ((x | 0) + (y | 0) * w) * 4;
const idxB = ((x + 1 | 0) + (y | 0) * w) * 4;
const idxC = ((x + 1 | 0) + (y + 1 | 0) * w) * 4;
const idxD = ((x | 0) + (y + 1 | 0) * w) * 4;
// Get channel data using sRGB approximation
const r1 = pixels[idxA] ** 2.2;
const r2 = pixels[idxB] ** 2.2;
const r3 = pixels[idxC] ** 2.2;
const r4 = pixels[idxD] ** 2.2;
const g1 = pixels[idxA + 1] ** 2.2;
const g2 = pixels[idxB + 1] ** 2.2;
const g3 = pixels[idxC + 1] ** 2.2;
const g4 = pixels[idxD + 1] ** 2.2;
const b1 = pixels[idxA + 2] ** 2.2;
const b2 = pixels[idxB + 2] ** 2.2;
const b3 = pixels[idxC + 2] ** 2.2;
const b4 = pixels[idxD + 2] ** 2.2;
// find value at location via linear interpolation
const xf = x % 1;
const yf = y % 1;
const rr = (r2 - r1) * xf + r1;
const gg = (g2 - g1) * xf + g1;
const bb = (b2 - b1) * xf + b1;
/// store channels as uncompressed sRGB
values.push((((r3 - r4) * xf + r4) - rr) * yf + rr);
values.push((((g3 - g4) * xf + g4) - gg) * yf + gg);
values.push((((b3 - b4) * xf + b4) - bb) * yf + bb);
} else {
// outside image
values.push(0,0);
}
x += nx;
y += ny;
}
values._nx = nx;
values._ny = ny;
values._x = x1;
values._y = y1;
return values;
}
function convertToMean(values) {
var i = 0,max = 0,v;
const results = [];
while (i < values.length) {
results.push(v = (values[i++] * 0.299 + values[i++] * 0.587 + values[i++] * 0.114) ** (1/2.2));
max = Math.max(v,max);
}
results._max = max;
results._nx = values._nx;
results._ny = values._ny;
results._x = values._x;
results._y = values._y;
return results;
}
function plotValues(ctx,values) {
const count = values.length;
const scaleX = ctx.canvas.width / count;
// not using max in example
// const scaleY = (ctx.canvas.height-3) / values._max;
const scaleY = (ctx.canvas.height-3) / 255;
ctx1.clearRect(0,ctx.canvas.height);
var i = 0;
ctx.beginPath();
ctx.strokeStyle = "#000";
ctx.lineWidth = 2;
while (i < count) {
const y = ctx.canvas.height - values[i] * scaleY + 1;
ctx.lineTo(i++ * scaleX,y);
}
ctx.stroke();
if (!mouse.button && mouse.overPlot) {
ctx.fillStyle = "#f008";
ctx.fillRect(mouse.px,1,ctx.canvas.height);
const val = values[mouse.px / scaleX | 0];
info.textContent = "Value: " + (val !== undefined ? val.toFixed(2) : "");
}
}
function update() {
ctx.clearRect(0,ctx.canvas.height);
ctx.drawImage(img,img.width * SCALE_IMAGE,img.height * SCALE_IMAGE);
var atSample = 0;
if (!mouse.button) {
if (line.canUse) {
if (line.haveData && mouse.overPlot) {
const count = line.data.length;
const scaleX = ctx1.canvas.width / count
atSample = mouse.px / scaleX;
}
}
}
if (mouse.button) {
if (mouse.dragging === 1) { // dragging line
line.x2 = mouse.ix;
line.y2 = mouse.iy;
line.canUse = true;
line.haveData = false;
} else if(mouse.overImage) {
mouse.dragging = 1;
line.x1 = mouse.ix;
line.y1 = mouse.iy;
line.canUse = false;
line.haveData = false;
canvas.style.cursor = "none";
}
} else {
mouse.dragging = 0;
canvas.style.cursor = "crosshair";
}
if (line.canUse) {
ctx.strokeStyle = "#F00";
ctx.strokeWidth = 2;
ctx.beginPath();
ctx.lineTo(line.x1,line.y1);
ctx.lineTo(line.x2,line.y2);
ctx.stroke();
if (atSample) {
ctx.fillStyle = "#FF0";
ctx.beginPath();
ctx.arc(
(line.data._x + line.data._nx * atSample) * SCALE_IMAGE,(line.data._y + line.data._ny * atSample) * SCALE_IMAGE,line.data[atSample | 0] / 32,Math.PI * 2
);
ctx.fill();
}
if (!line.haveData) {
const vals = getProfile(imgData,line.x1,line.y1,line.x2,line.y2,1);
line.data = convertToMean(vals);
line.haveData = true;
plotValues(ctx1,line.data);
} else {
plotValues(ctx1,line.data);
}
}
requestAnimationFrame(update);
}
function mouseEvents(e){
if (bounds) {
mouse.ix = e.pageX - bounds.left;
mouse.iy = e.pageY - bounds.top;
mouse.overImage = mouse.ix >= 0 && mouse.ix < bounds.width && mouse.iy >= 0 && mouse.iy < bounds.height;
mouse.px = e.pageX - bounds1.left;
mouse.py = e.pageY - bounds1.top;
mouse.overPlot = mouse.px >= 0 && mouse.px < bounds1.width && mouse.py >= 0 && mouse.py < bounds1.height;
}
mouse.button = e.type === "mousedown" ? true : e.type === "mouseup" ? false : mouse.button;
}
canvas {
border: 2px solid black;
}
<canvas id="canvas"></canvas>
<div id="info">Click drag line over image</div>
<canvas id="canvas1"></canvas>
图片来源:https://commons.wikimedia.org/w/index.php?curid=93680693,作者:BethGuay-自己的作品,CC BY-SA 4.0,
版权声明:本文内容由互联网用户自发贡献,该文观点与技术仅代表作者本人。本站仅提供信息存储空间服务,不拥有所有权,不承担相关法律责任。如发现本站有涉嫌侵权/违法违规的内容, 请发送邮件至 dio@foxmail.com 举报,一经查实,本站将立刻删除。