cars.applications.resampling.bicubic_resampling

this module contains the dense_matching application class.

Module Contents

Classes

BicubicResampling

BicubicResampling

Functions

generate_epipolar_images_wrapper(→ Dict[str, ...)	Compute disparity maps from image objects. This function will be run
get_sensors_bounds(sensor_image_left, sensor_image_right)	Get bounds of sensor images
check_tiles_in_sensor(sensor_image_left, ...)	Check if epipolar tiles will be used.
check_tile_inclusion(left_sensor_bounds, ...)	Check if tile is in sensor image

class cars.applications.resampling.bicubic_resampling.BicubicResampling(conf=None)

Bases: cars.applications.resampling.resampling.Resampling

BicubicResampling

check_conf(conf)

Check configuration

Parameters

conf (dict) – configuration to check

Returns

overloaded configuration

Return type

dict

pre_run(grid_left, tile_width, tile_height)

Pre run some computations : tiling grid

Parameters

• grid_left (CarsDataset) – left grid

• optimum_tile_size (int) – optimum tile size

Returns

epipolar_regions_grid, epipolar_regions, opt_epipolar_tile_size, largest_epipolar_region,

run(sensor_image_left, sensor_image_right, grid_left, grid_right, orchestrator=None, pair_folder=None, pair_key='PAIR_0', margins_fun=None, tile_width=None, tile_height=None, add_color=True, add_classif=True, epipolar_roi=None)

Run resampling application.

Creates left and right CarsDataset filled with xarray.Dataset, corresponding to sensor images resampled in epipolar geometry.

Parameters

• sensor_images_left (CarsDataset) – tiled sensor left image Dict Must contain keys : “image”, “color”, “geomodel”, “no_data”, “mask”, “classification”. Paths must be absolutes

• sensor_images_right (CarsDataset) – tiled sensor right image Dict Must contain keys : “image”, “color”, “geomodel”, “no_data”, “mask”, “classification”. Paths must be absolutes

• grid_left –

  left epipolar grid Grid CarsDataset contains :

  • A single tile stored in [0,0], containing a (N, M, 2) shape

    array in xarray Dataset

  • Attributes containing: “grid_spacing”, “grid_origin”, “epipolar_size_x”, “epipolar_size_y”, “epipolar_origin_x”, “epipolar_origin_y”, epipolar_spacing_x”, “epipolar_spacing”, “disp_to_alt_ratio”, :type grid_left: CarsDataset

• grid_right (CarsDataset) –

  right epipolar grid. Grid CarsDataset contains :

  • A single tile stored in [0,0], containing a (N, M, 2) shape array in xarray Dataset

  • Attributes containing: “grid_spacing”, “grid_origin”, “epipolar_size_x”, “epipolar_size_y”, “epipolar_origin_x”, “epipolar_origin_y”, epipolar_spacing_x”, “epipolar_spacing”, “disp_to_alt_ratio”,

• orchestrator – orchestrator used

• pair_folder (directory to save files to) – folder used for current pair

• pair_key (str) – pair id

• margins_fun (fun) – margins function to use

• optimum_tile_size (int) – optimum tile size to use

• tile_width (int) – width of tile

• tile_height (int) – height of tile

• add_color (bool) – add color image to dataset

• add_classif (bool) – add classif to dataset

• epipolar_roi (list(int), [row_min, row_max, col_min, col_max]) – Epipolar roi to use if set. Set None tiles outsize roi

Returns

left epipolar image, right epipolar image. Each CarsDataset contains:

• N x M Delayed tiles. Each tile will be a future xarray Dataset containing:

  • data with keys : “im”, “msk”, “color”, “classif”

  • attrs with keys: “margins” with “disp_min” and “disp_max” “transform”, “crs”, “valid_pixels”, “no_data_mask”,

    ”no_data_img”

• attributes containing: “largest_epipolar_region”,”opt_epipolar_tile_size”,

  ”disp_min_tiling”, “disp_max_tiling”

Return type

Tuple(CarsDataset, CarsDataset)

cars.applications.resampling.bicubic_resampling.generate_epipolar_images_wrapper(left_overlaps, right_overlaps, window, epipolar_size_x, epipolar_size_y, img1, img2, grid1, grid2, interpolator_image, interpolator_color, interpolator_classif, interpolator_mask, step=None, used_disp_min=None, used_disp_max=None, add_color=True, add_classif=True, color1=None, mask1=None, mask2=None, classif1=None, classif2=None, nodata1=0, nodata2=0, saving_info_left=None, saving_info_right=None) → Dict[str, Tuple[xarray.Dataset, xarray.Dataset]]

Compute disparity maps from image objects. This function will be run as a delayed task. If user want to correctly save dataset, the user must provide saving_info_left and right. See cars_dataset.fill_dataset.

Parameters

• left_overlaps (dict) – Overlaps of left image, with row_min, row_max, col_min and col_max keys.

• right_overlaps (dict) – Overlaps of right image, with row_min, row_max, col_min and col_max keys.

• window (dict) – Window considered in generation, with row_min, row_max, col_min and col_max keys.

Returns

Left image object, Right image object (if exists)

Returned objects are composed of dataset with :

• cst.EPI_IMAGE

• cst.EPI_MSK (if given)

• cst.EPI_COLOR (for left, if given)

cars.applications.resampling.bicubic_resampling.get_sensors_bounds(sensor_image_left, sensor_image_right)

Get bounds of sensor images Bounds: BoundingBox(left, bottom, right, top)

Parameters

• sensor_image_left (dict) – left sensor

• sensor_image_right (dict) – right sensor

Returns

left image bounds, right image bounds

Return type

tuple(list, list)

cars.applications.resampling.bicubic_resampling.check_tiles_in_sensor(sensor_image_left, sensor_image_right, image_tiling, grid_left, grid_right)

Check if epipolar tiles will be used. A tile is not used if is outside sensor bounds

Parameters

• sensor_image_left (dict) – left sensor

• sensor_image_right (dict) – right sensor

• image_tiling (np.array) – epipolar tiling grid

• grid_left (CarsDataset) – left epipolar grid

• grid_right (CarsDataset) – right epipolar grid

Returns

left in sensor, right in sensor

Return type

np.array(bool), np.array(bool)

cars.applications.resampling.bicubic_resampling.check_tile_inclusion(left_sensor_bounds, right_sensor_bounds, sensor_pos_left, sensor_pos_right)

Check if tile is in sensor image

Parameters

• left_sensor_bounds (list) – bounds of left sensor

• right_sensor_bounds (list) – bounds of right sensor

• sensor_pos_left (np.array) – left sensor position

• sensor_pos_right (np.array) – right sensor position

Returns

left tile in sensor image left, right tile in sensor image right

Return type

tuple(bool, bool)