static bool addColorLayerFromImage(const cv::Mat& image, const std::string& layer,
grid_map::GridMap& gridMap)
{
if (gridMap.getSize()(0) != image.rows || gridMap.getSize()(1) != image.cols) {
std::cerr << "Image size does not correspond to grid map size!" << std::endl;
return false;
}
bool hasAlpha = false;
if (image.channels() >= 4) hasAlpha = true;
cv::Mat imageRGB;
if (hasAlpha) {
cv::cvtColor(image, imageRGB, CV_BGRA2RGB);
} else {
imageRGB = image;
}
gridMap.add(layer);
for (GridMapIterator iterator(gridMap); !iterator.isPastEnd(); ++iterator) {
const auto& cvColor = imageRGB.at<cv::Vec<Type_, 3>>((*iterator)(0), (*iterator)(1));
Eigen::Vector3i colorVector;
colorVector(0) = cvColor[0];
colorVector(1) = cvColor[1];
colorVector(2) = cvColor[2];
colorVectorToValue(colorVector, gridMap.at(layer, *iterator));
}
return true;
}
bool GridMapRosConverter::addColorLayerFromImage(const sensor_msgs::Image& image,
const std::string& layer,
grid_map::GridMap& gridMap)
{
cv_bridge::CvImagePtr cvPtr;
try {
cvPtr = cv_bridge::toCvCopy(image, image.encoding);
// cvPtr = cv_bridge::toCvCopy(image, sensor_msgs::image_encodings::BGR8); // FixMe
} catch (cv_bridge::Exception& e) {
ROS_ERROR("cv_bridge exception: %s", e.what());
return false;
}
gridMap.add(layer);
if (gridMap.getSize()(0) != image.height || gridMap.getSize()(1) != image.width) {
ROS_ERROR("Image size does not correspond to grid map size!");
return false;
}
for (GridMapIterator iterator(gridMap); !iterator.isPastEnd(); ++iterator) {
const auto& cvColor = cvPtr->image.at<cv::Vec3b>((*iterator)(0), (*iterator)(1));
Eigen::Vector3i colorVector;
// TODO Add cases for different image encodings.
colorVector(2) = cvColor[0]; // From BGR to RGB.
colorVector(1) = cvColor[1];
colorVector(0) = cvColor[2];
colorVectorToValue(colorVector, gridMap.at(layer, *iterator));
}
return true;
}
void ElevationChangeDetection::computeElevationChange(grid_map::GridMap& elevationMap)
{
elevationMap.add("elevation_change", elevationMap.get(layer_));
elevationMap.clear("elevation_change");
for (GridMapIterator iterator(elevationMap);
!iterator.isPastEnd(); ++iterator) {
// Check if elevation map has valid value
if (!elevationMap.isValid(*iterator, layer_)) continue;
double height = elevationMap.at(layer_, *iterator);
// Get the ground truth height
Vector2d position, groundTruthPosition;
Array2i groundTruthIndex;
elevationMap.getPosition(*iterator, position);
groundTruthMap_.getIndex(position, groundTruthIndex);
if (!groundTruthMap_.isValid(groundTruthIndex, layer_)) continue;
double groundTruthHeight = groundTruthMap_.at(layer_, groundTruthIndex);
// Add to elevation change map
double diffElevation = std::abs(height - groundTruthHeight);
if (diffElevation <= threshold_) continue;
elevationMap.at("elevation_change", *iterator) = diffElevation;
}
}
void FillPolygonAreas(grid_map::GridMap &out_grid_map, const std::vector<std::vector<geometry_msgs::Point>> &in_area_points,
const std::string &in_grid_layer_name, const int in_layer_background_value,
const int in_layer_min_value, const int in_fill_color, const int in_layer_max_value,
const std::string &in_tf_target_frame, const std::string &in_tf_source_frame,
const tf::TransformListener &in_tf_listener)
{
if(!out_grid_map.exists(in_grid_layer_name))
{
out_grid_map.add(in_grid_layer_name);
}
out_grid_map[in_grid_layer_name].setConstant(in_layer_background_value);
cv::Mat original_image;
grid_map::GridMapCvConverter::toImage<unsigned char, 1>(out_grid_map,
in_grid_layer_name,
CV_8UC1,
in_layer_min_value,
in_layer_max_value,
original_image);
cv::Mat filled_image = original_image.clone();
tf::StampedTransform tf = FindTransform(in_tf_target_frame, in_tf_source_frame, in_tf_listener);
// calculate out_grid_map position
grid_map::Position map_pos = out_grid_map.getPosition();
double origin_x_offset = out_grid_map.getLength().x() / 2.0 - map_pos.x();
double origin_y_offset = out_grid_map.getLength().y() / 2.0 - map_pos.y();
for (const auto &points : in_area_points)
{
std::vector<cv::Point> cv_points;
for (const auto &p : points)
{
// transform to GridMap coordinate
geometry_msgs::Point tf_point = TransformPoint(p, tf);
// coordinate conversion for cv image
double cv_x = (out_grid_map.getLength().y() - origin_y_offset - tf_point.y) / out_grid_map.getResolution();
double cv_y = (out_grid_map.getLength().x() - origin_x_offset - tf_point.x) / out_grid_map.getResolution();
cv_points.emplace_back(cv::Point(cv_x, cv_y));
}
cv::fillConvexPoly(filled_image, cv_points.data(), cv_points.size(), cv::Scalar(in_fill_color));
}
// convert to ROS msg
grid_map::GridMapCvConverter::addLayerFromImage<unsigned char, 1>(filled_image,
in_grid_layer_name,
out_grid_map,
in_layer_min_value,
in_layer_max_value);
}
static bool addLayerFromImage(const cv::Mat& image, const std::string& layer,
grid_map::GridMap& gridMap, const float lowerValue = 0.0,
const float upperValue = 1.0, const double alphaThreshold = 0.5)
{
if (gridMap.getSize()(0) != image.rows || gridMap.getSize()(1) != image.cols) {
std::cerr << "Image size does not correspond to grid map size!" << std::endl;
return false;
}
bool isColor = false;
if (image.channels() >= 3) isColor = true;
bool hasAlpha = false;
if (image.channels() >= 4) hasAlpha = true;
cv::Mat imageMono;
if (isColor && !hasAlpha) {
cv::cvtColor(image, imageMono, CV_BGR2GRAY);
} else if (isColor && hasAlpha) {
cv::cvtColor(image, imageMono, CV_BGRA2GRAY);
} else if (!isColor && !hasAlpha){
imageMono = image;
} else {
std::cerr << "Something went wrong when adding grid map layer form image!" << std::endl;
return false;
}
const float mapValueDifference = upperValue - lowerValue;
const float maxImageValue = (float)std::numeric_limits<Type_>::max();
const Type_ alphaTreshold = (Type_)(alphaThreshold * std::numeric_limits<Type_>::max());
gridMap.add(layer);
grid_map::Matrix& data = gridMap[layer];
for (GridMapIterator iterator(gridMap); !iterator.isPastEnd(); ++iterator) {
const Index index(*iterator);
// Check for alpha layer.
if (hasAlpha) {
const Type_ alpha = image.at<cv::Vec<Type_, NChannels_>>(index(0), index(1))[NChannels_ - 1];
if (alpha < alphaTreshold) continue;
}
// Compute value.
const Type_ imageValue = imageMono.at<Type_>(index(0), index(1));
const float mapValue = lowerValue + mapValueDifference * ((float) imageValue / maxImageValue);
data(index(0), index(1)) = mapValue;
}
return true;
};
bool GridMapRosConverter::fromOccupancyGrid(const nav_msgs::OccupancyGrid& occupancyGrid,
const std::string& layer, grid_map::GridMap& gridMap)
{
const Size size(occupancyGrid.info.width, occupancyGrid.info.height);
const double resolution = occupancyGrid.info.resolution;
const Length length = resolution * size.cast<double>();
const string& frameId = occupancyGrid.header.frame_id;
Position position(occupancyGrid.info.origin.position.x, occupancyGrid.info.origin.position.y);
// Different conventions of center of map.
position += 0.5 * length.matrix();
const auto& orientation = occupancyGrid.info.origin.orientation;
if (orientation.w != 1.0 && !(orientation.x == 0 && orientation.y == 0
&& orientation.z == 0 && orientation.w == 0)) {
ROS_WARN_STREAM("Conversion of occupancy grid: Grid maps do not support orientation.");
ROS_INFO_STREAM("Orientation of occupancy grid: " << endl << occupancyGrid.info.origin.orientation);
return false;
}
if (size.prod() != occupancyGrid.data.size()) {
ROS_WARN_STREAM("Conversion of occupancy grid: Size of data does not correspond to width * height.");
return false;
}
if ((gridMap.getSize() != size).any() || gridMap.getResolution() != resolution
|| (gridMap.getLength() != length).any() || gridMap.getPosition() != position
|| gridMap.getFrameId() != frameId || !gridMap.getStartIndex().isZero()) {
gridMap.setTimestamp(occupancyGrid.header.stamp.toNSec());
gridMap.setFrameId(frameId);
gridMap.setGeometry(length, resolution, position);
}
// Reverse iteration is required because of different conventions
// between occupancy grid and grid map.
grid_map::Matrix data(size(0), size(1));
for (std::vector<int8_t>::const_reverse_iterator iterator = occupancyGrid.data.rbegin();
iterator != occupancyGrid.data.rend(); ++iterator) {
size_t i = std::distance(occupancyGrid.data.rbegin(), iterator);
data(i) = *iterator != -1 ? *iterator : NAN;
}
gridMap.add(layer, data);
return true;
}
bool GridMapRosConverter::fromMessage(const grid_map_msgs::GridMap& message, grid_map::GridMap& gridMap)
{
gridMap.setTimestamp(message.info.header.stamp.toNSec());
gridMap.setFrameId(message.info.header.frame_id);
gridMap.setGeometry(Length(message.info.length_x, message.info.length_y), message.info.resolution,
Position(message.info.pose.position.x, message.info.pose.position.y));
if (message.layers.size() != message.data.size()) {
ROS_ERROR("Different number of layers and data in grid map message.");
return false;
}
for (unsigned int i = 0; i < message.layers.size(); i++) {
Matrix data;
multiArrayMessageCopyToMatrixEigen(message.data[i], data); // TODO Could we use the data mapping (instead of copying) method here?
// TODO Check if size is good. size_ << getRows(message.data[0]), getCols(message.data[0]);
gridMap.add(message.layers[i], data);
}
gridMap.setBasicLayers(message.basic_layers);
gridMap.setStartIndex(Index(message.outer_start_index, message.inner_start_index));
return true;
}
bool GridMapRosConverter::addLayerFromImage(const sensor_msgs::Image& image,
const std::string& layer, grid_map::GridMap& gridMap,
const double lowerValue, const double upperValue)
{
cv_bridge::CvImagePtr cvPtrAlpha, cvPtrMono;
// If alpha channel exist, read it.
if (image.encoding == sensor_msgs::image_encodings::BGRA8
|| image.encoding == sensor_msgs::image_encodings::BGRA16) {
try {
cvPtrAlpha = cv_bridge::toCvCopy(image, image.encoding);
} catch (cv_bridge::Exception& e) {
ROS_ERROR("cv_bridge exception: %s", e.what());
return false;
}
}
unsigned int depth;
// Convert color image to grayscale.
try {
if (image.encoding == sensor_msgs::image_encodings::BGRA8
|| image.encoding == sensor_msgs::image_encodings::BGR8
|| image.encoding == sensor_msgs::image_encodings::MONO8) {
cvPtrMono = cv_bridge::toCvCopy(image,
sensor_msgs::image_encodings::MONO8);
depth = std::pow(2, 8);
ROS_DEBUG("Color image converted to mono8");
} else if (image.encoding == sensor_msgs::image_encodings::BGRA16
|| image.encoding == sensor_msgs::image_encodings::BGR16
|| image.encoding == sensor_msgs::image_encodings::MONO16) {
cvPtrMono = cv_bridge::toCvCopy(image,
sensor_msgs::image_encodings::MONO16);
depth = std::pow(2, 16);
ROS_DEBUG("Color image converted to mono16");
} else {
ROS_ERROR("Expected BGR, BGRA, or MONO image encoding.");
return false;
}
} catch (cv_bridge::Exception& e) {
ROS_ERROR("cv_bridge exception: %s", e.what());
return false;
}
gridMap.add(layer);
if (gridMap.getSize()(0) != image.height
|| gridMap.getSize()(1) != image.width) {
ROS_ERROR("Image size does not correspond to grid map size!");
return false;
}
for (GridMapIterator iterator(gridMap); !iterator.isPastEnd(); ++iterator) {
// Set transparent values.
if (image.encoding == sensor_msgs::image_encodings::BGRA8) {
const auto& cvAlpha = cvPtrAlpha->image.at<cv::Vec4b>((*iterator)(0),
(*iterator)(1));
unsigned int alpha = cvAlpha[3];
if (cvAlpha[3] < depth / 2)
continue;
}
if (image.encoding == sensor_msgs::image_encodings::BGRA16) {
const auto& cvAlpha = cvPtrAlpha->image.at<cv::Vec<uchar, 8>>(
(*iterator)(0), (*iterator)(1));
int alpha = (cvAlpha[6] << 8) + cvAlpha[7];
if (alpha < depth / 2)
continue;
}
// Compute height.
unsigned int grayValue;
if (depth == std::pow(2, 8)) {
uchar cvGrayscale = cvPtrMono->image.at<uchar>((*iterator)(0),
(*iterator)(1));
grayValue = cvGrayscale;
}
if (depth == std::pow(2, 16)) {
const auto& cvGrayscale = cvPtrMono->image.at<cv::Vec2b>((*iterator)(0),
(*iterator)(1));
grayValue = (cvGrayscale[0] << 8) + cvGrayscale[1];
}
double height = lowerValue
+ (upperValue - lowerValue) * ((double) grayValue / (double) depth);
gridMap.at(layer, *iterator) = height;
}
return true;
}
