corner_responseCornerResponseCornerResponsecorner_responsecorner_response🔗

Short description🔗

corner_responseCornerResponseCornerResponsecorner_responsecorner_response — Searching corners in images.

Signature🔗

corner_response( image Image, out image ImageCorner, integer Size, real Weight )

Description🔗

The operator corner_responseCornerResponse extracts gray value corners in an image. The formula for the calculation of the response is:

\[\begin{eqnarray*} R(x,y) &=& A(x,y) \cdot B(x,y) - C^2(x,y) - Weight \cdot \left(A(x,y) + B(x,y)\right)^2\\ A(x,y) &=& W(u,v) \ast (\nabla_{x} I(x,y))^2\\ B(x,y) &=& W(u,v) \ast (\nabla_{y} I(x,y))^2\\ C(c,y) &=& W(u,v) \ast \left(\nabla_{x} I(x,y) \nabla_{y} I(x,y)\right) \end{eqnarray*}\]

with \(I\): input image, \(R\): output image of the filter, \(A\), and \(B\): smoothed directional derivatives, \(C\): smoothing of the products of the directional derivatives. The operator gauss_imageGaussImage is used for smoothing (\(W\)), the operator sobel_ampSobelAmp is used for calculating the derivative \(\nabla\)).

The corner response function is invariant with regard to rotation. In order to achieve a suitable dependency of the function \(R(x,y)\) on the local gradient, the parameter Weightweightweight must be set to 0.04. With this, only gray value corners will return positive values for \(R(x,y)\), while straight edges will receive negative values. The output image type is identical to the input image type. Therefore, the negative output values are set to 0 if byte images are used as input images. If this is not desired, the input image should be converted into a real or int2 image with convert_image_typeConvertImageType.

Attention🔗

Note that filter operators may return unexpected results if an image with a reduced domain is used as input. Please refer to the chapter Filters.

Execution information🔗

Execution information

Multithreading type: reentrant (runs in parallel with non-exclusive operators).
Multithreading scope: global (may be called from any thread).
Automatically parallelized on tuple level.
Automatically parallelized on channel level.
Automatically parallelized on domain level.

Parameters🔗

Imageimageimage (input_object) (multichannel-)image(-array) → object (byte / uint2 / int2 / real)HObject (byte / uint2 / int2 / real)HImage (byte / uint2 / int2 / real)HObject (byte / uint2 / int2 / real)Hobject (byte / uint2 / int2 / real)

Input image.

ImageCornerimageCornerimage_corner (output_object) multichannel-image(-array) → object (byte / uint2 / int2 / real)HObject (byte / uint2 / int2 / real)HImage (byte / uint2 / int2 / real)HObject (byte / uint2 / int2 / real)Hobject * (byte / uint2 / int2 / real)

Result of the filtering.

Number of elements: ImageCorner == Image

Sizesizesize (input_control) integer → (integer)HTuple (Hlong)HTuple (int / long)intHtuple (Hlong)

Desired filtersize of the graymask.

Default: 33
Suggested values: 3, 5, 7, 9, 113, 5, 7, 9, 11

Weightweightweight (input_control) real → (real)HTuple (double)HTuple (double)floatHtuple (double)

Weighting.

Default: 0.040.04
Value range: 0.0 ≤ Weight ≤ 0.3
Minimum increment: 0.001
Recommended increment: 0.01

Example🔗

(C)

read_image(&Fabrik,"fabrik")\;
corner_response(Fabrik,&CornerResponse,3,0.04)\;
local_max(CornerResponse,&LocalMax)\;
disp_image(Fabrik,WindowHandle)\;
set_color(WindowHandle,"red")\;
disp_region(LocalMax,WindowHandle)\;

Combinations with other operators🔗

Combinations

Possible successors

local_maxLocalMax, thresholdThreshold

References🔗

C.G. Harris, M.J. Stephens, “A combined corner and edge detector”’; Proc. of the 4th Alvey Vision Conference; August 1988; pp. 147-152.

H. Breit, ``Bestimmung der Kameraeigenbewegung und Gewinnung von Tiefendaten aus monokularen Bildfolgen’‘; Diplomarbeit am Lehrstuhl für Nachrichtentechnik der TU München; 30. September 1990.

Module🔗

Foundation