前言
作业比较简单,关键在于计算切割点算法实现
作业要求
非提高题主要还是要实现这个算法
De Cassteljau算法原理
老师的pdf已经把步骤写得很清楚了,这里再附上一张图方便理解
作业具体代码
我个人感觉这个算法用队列实现还是比较清晰的
cv::Point2f recursive_bezier(const vector<cv::Point2f> &control_points, float t)
{
// TODO: Implement de Casteljau's algorithm
queue<cv::Point2f> q; //用队列更好处理边界
for (auto i : control_points) {
q.push(i);
}
while (q.size() > 1) {
int n = q.size();
for (int i = 1; i < n; ++i) { //n个点计算出n-1条线上的点
cv::Point2f a = q.front(); q.pop();
cv::Point2f b = q.front();
if (i == n - 1) q.pop();
cv::Point2f c = (1 - t) * a + t * b;
q.push(c);
}
}
return q.front();
}
void bezier(const std::vector<cv::Point2f> &control_points, cv::Mat &window)
{
for (double t = 0.0; t <= 1.0; t += 0.001)
{
cv::Point2f point = recursive_bezier(control_points, t);
window.at<cv::Vec3b>(point.y, point.x)[1] = 255;
}
}
需要注意老师要求是画两条线重合为黄色,naive_bezier实现时是使用通道3也就是索引为2的通道赋值255,这就是红色,如果改为索引1就是绿色,索引0就是蓝色,这其实就是三原色,绿色和红色混合就是黄色
提高题
本人比较菜,感觉双线性插值比较麻烦,采用这位大佬的类似SSAA方法
games101 作业4提高部分
这个方法主要是采用距离归一化的方法,区分点和中心像素点,真正计算时每次都是取到第二行第二列的中心点,所以for循环是从[-1,1],正好通过中心点遍历所有周围的像素点
扩大到3x3的范围想必也是为了减少误差
代码
void bezier(const vector<cv::Point2f> &control_points, cv::Mat &window)
{
// TODO: Iterate through all t = 0 to t = 1 with small steps, and call de Casteljau's
// recursive Bezier algorithm.
for (double t = 0.0; t <= 1.0; t += 0.001)
{
auto point = recursive_bezier(control_points, t); //求最终点
auto centerPoint = cv::Point2f(int(point.x) + 0.5f, int(point.y) + 0.5f);
auto x = point.x, y = point.y, centerx = centerPoint.x, centery = centerPoint.y;
for (int i = -1; i <= 1; ++i) {
for (int j = -1; j <= 1; ++j) {
float dis = sqrt(pow(x - (centerx - i), 2) + pow(y - (centery - j), 2)); //勾股定理求距离
float nor_dis = dis * sqrt(2) / 3; //归一化,由于最大距离是3根号2/2,所以乘以根号2/3就是1,这样就能将颜色最大限制在255
float color = 255.0f * (1 - nor_dis); //距离越近,颜色越深
if (window.at<cv::Vec3b>(centery - j, centerx - i)[2] < color)
window.at<cv::Vec3b>(centery - j, centerx - i)[2] = color;
}
}
}
}
void naive_bezier(const vector<cv::Point2f> &points, cv::Mat &window)
{
auto &p_0 = points[0];
auto &p_1 = points[1];
auto &p_2 = points[2];
auto &p_3 = points[3];
for (double t = 0.0; t <= 1.0; t += 0.001)
{
auto point = pow(1 - t, 3) * p_0 + 3 * t * pow(1 - t, 2) * p_1 +
3 * pow(t, 2) * (1 - t) * p_2 + pow(t, 3) * p_3;
auto centerPoint = cv::Point2f(int(point.x) + 0.5f, int(point.y) + 0.5f);
auto x = point.x, y = point.y, centerx = centerPoint.x, centery = centerPoint.y;
for (int i = -1; i <= 1; ++i) {
for (int j = -1; j <= 1; ++j) {
float dis = sqrt(pow(x - (centerx - i), 2) + pow(y - (centery - j), 2)); //勾股定理求距离
float nor_dis = dis * sqrt(2) / 3;
float color = 255.0f * (1 - nor_dis);
if (window.at<cv::Vec3b>(centery - j, centerx - i)[1] < color)
window.at<cv::Vec3b>(centery - j, centerx - i)[1] = color;
}
}
}
}
总结
最简单的一次作业,难点在于提高题的锯齿优化,但思路还是很明确的,实现有点东西