Open Source Computer Vision Library https://opencv.org/
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
 
 

91 lines
2.7 KiB

// This file is part of OpenCV project.
// It is subject to the license terms in the LICENSE file found in the top-level directory
// of this distribution and at http://opencv.org/license.html.
if (typeof module !== 'undefined' && module.exports) {
// The environment is Node.js
var cv = require('./opencv.js'); // eslint-disable-line no-var
}
QUnit.module('Camera Calibration and 3D Reconstruction', {});
QUnit.test('constants', function(assert) {
assert.strictEqual(typeof cv.LMEDS, 'number');
assert.strictEqual(typeof cv.RANSAC, 'number');
assert.strictEqual(typeof cv.RHO, 'number');
});
QUnit.test('findHomography', function(assert) {
let srcPoints = cv.matFromArray(4, 1, cv.CV_32FC2, [
56,
65,
368,
52,
28,
387,
389,
390,
]);
let dstPoints = cv.matFromArray(4, 1, cv.CV_32FC2, [
0,
0,
300,
0,
0,
300,
300,
300,
]);
const mat = cv.findHomography(srcPoints, dstPoints);
assert.ok(mat instanceof cv.Mat);
});
QUnit.test('Rodrigues', function(assert) {
// Converts a rotation matrix to a rotation vector and vice versa
// data64F is the output array
const rvec0 = cv.matFromArray(1, 3, cv.CV_64F, [1,1,1]);
let rMat0 = new cv.Mat();
let rvec1 = new cv.Mat();
// Args: input Mat, output Mat. The function mutates the output Mat, so the function does not return anything.
// cv.Rodrigues (InputArray=src, OutputArray=dst, jacobian=0)
// https://docs.opencv.org/2.4/modules/calib3d/doc/camera_calibration_and_3d_reconstruction.html#void%20Rodrigues(InputArray%20src,%20OutputArray%20dst,%20OutputArray%20jacobian)
// vec to Mat, starting number is 3 long and each element is 1.
cv.Rodrigues(rvec0, rMat0);
assert.ok(rMat0.data64F.length == 9);
assert.ok(0.23 > rMat0.data64F[0] > 0.22);
// convert Mat to Vec, should be same as what we started with, 3 long and each item should be a 1.
cv.Rodrigues(rMat0, rvec1);
assert.ok(rvec1.data64F.length == 3);
assert.ok(1.01 > rvec1.data64F[0] > 0.9);
// Answer should be around 1: 0.9999999999999999
});
QUnit.test('estimateAffine2D', function(assert) {
const inputs = cv.matFromArray(4, 1, cv.CV_32FC2, [
1, 1,
80, 0,
0, 80,
80, 80
]);
const outputs = cv.matFromArray(4, 1, cv.CV_32FC2, [
21, 51,
70, 77,
40, 40,
10, 70
]);
const M = cv.estimateAffine2D(inputs, outputs);
assert.ok(M instanceof cv.Mat);
assert.deepEqual(Array.from(M.data), [
23, 55, 97, 126, 87, 139, 227, 63, 0, 0,
0, 0, 0, 0, 232, 191, 71, 246, 12, 68,
165, 35, 53, 64, 99, 56, 27, 66, 14, 254,
212, 63, 103, 102, 102, 102, 102, 102, 182, 191,
195, 252, 174, 22, 55, 97, 73, 64
]);
});