test-canvas/image.js

106 lines
3.3 KiB
JavaScript
Raw Normal View History

2024-10-01 15:55:58 +00:00
const canvas = document.getElementById("canvas");
const ctx = canvas.getContext("2d");
const img = new Image();
img.crossOrigin = "anonymous";
img.src = "/rose.png";
img.onload = () => {
/*canvas.offscreenCanvas = document.createElement("canvas");
canvas.offscreenCanvas.height = img.height;
canvas.offscreenCanvas.width = img.width;
canvas.offscreenCanvas.getContext("2d").drawImage(img, 0, 0);
*/
ctx.drawImage(img, 0, 0);
};
function invert(imgData) {
const data = imgData.data;
for (let i = 0; i < data.length; i += 4) {
//const avg = (data[i] + data[i + 1] + data[i + 2]) / 3;
data[i] = 255 - data[i];
data[i+1] = 255 - data[i+1];
data[i+2] = 255 - data[i+2];
}
}
function greyscale(imgData) {
const data = imgData.data;
for (let i = 0; i < data.length; i += 4) {
const avg = (data[i] + data[i + 1] + data[i + 2]) / 3;
data[i] = avg;
data[i+1] = avg;
data[i+2] = avg;
}
}
function clamp(x, min, max) {
return Math.min(Math.max(x, min), max);
}
function convolve(imageData, kernel) {
const width = imageData.width;
const height = imageData.height;
const data = imageData.data;
const out = new ImageData(width, height);
const kernelSize = Math.sqrt(kernel.length); // Assuming kernel is square
const halfKernel = Math.floor(kernelSize / 2);
for (let y = 0; y < height; y++) {
for (let x = 0; x < width; x++) {
let r = 0, g = 0, b = 0;
for (let ky = -halfKernel; ky <= halfKernel; ky++) {
for (let kx = -halfKernel; kx <= halfKernel; kx++) {
const px = x + kx;
const py = y + ky;
if (px >= 0 && px < width && py >= 0 && py < height) {
const idx = (py * width + px) * 4;
const kernelValue = kernel[(ky + halfKernel) * kernelSize + (kx + halfKernel)];
r += data[idx] * kernelValue;
g += data[idx + 1] * kernelValue;
b += data[idx + 2] * kernelValue;
}
}
}
out.data[(y*width+x)*4] = r;
out.data[(y*width+x)*4 + 1] = g;
out.data[(y*width+x)*4 + 2] = b;
out.data[(y*width+x)*4 + 3] = data[(y*width+x) * 4 + 3]; // preserve original alpha
}
}
return out;
}
// multiply two 3x3 matrices
function matMult(a, b) {
const result = new Array(9).fill(0);
for (let j = 0; j < 3; ++j){
for (let i = 0; i < 3; ++i) {
for (let k = 0; k < 3; ++k) {
result[j*3+i] += a[j*3+k] * b[k*3+i];
}
}
}
return result;
}
canvas.onclick = () => {
const imgData = ctx.getImageData(0, 0, canvas.width, canvas.height);
ctx.clearRect(0, 0, canvas.height, canvas.width);
//invert(imgData);
//greyscale(imgData);
//const identityKernel = [0, 0, 0, 0, 1, 0, 0, 0, 0];
const gaussianKernel = [0.075, 0.124, 0.075, 0.124, 0.204, 0.124, 0.075, 0.124, 0.075]
const convolved = convolve(imgData, gaussianKernel);
ctx.putImageData(convolved, 0, 0);
};
const matrixA = [1, 2, 3, 4, 5, 6, 7, 8, 9];
const matrixB = [9, 8, 7, 6, 5, 4, 3, 2, 1];
const result = matMult(matrixA, matrixB);
console.log(result);