MV Convolution DirectShow Filter

The MontiVision Convolution filter is used to convolve an image with one or more user defined matrices. If more than one matrix is used, the matrices results can be combined using several different methods. The size of the matrices is fully variable between 1 X 1 and 7 X 7.

 

File: convolution.ax
CLSID: CLSID_MVConvolution

 

Supported Interfaces

IMVConvolution

IMVHistogram

IMVHistogram2

IMVHistogram3

IMVRegionOfInterest

IMVFilterData

IMVAbout

IMVFilterInfo

 

Supported COM Events

The filter supports following events through the IMVFilterEvents interface. Look at the filter events page for more information.

Event

Value

Description
MV_ALGORITHM_START_EVENT

0

A filter starts processing a video frame or sample.
MV_ALGORITHM_END_EVENT

1

A filter finishes processing a video frame or sample.

 

MV Convolution Property Page

Uses the IMVConvolution interface

MontiVision Video Convolution DirectShow Filter Property Page
Convolution filter property page

Number of kernels Specifies the number of kernels convolved with the current image.
Selected kernel Used to select the kernel to be modified. The size of the kernel is changed using the row and column radio boxes. The kernel numerals are modified by changing the values in the edit controls. The kernel is updated after the apply button is clicked.
Combination

Select the method to be used to combine the results of the convolutions.

Right shift pixel Used to scale the kernels convolution results. Every pixel channel convolution result value is divided by 2"Right shift pixel".
Anchor The [x, y] coordinates of the anchor cell in the kernel. The top left corner would be [0, 0] and the bottom right corner would be [Columns-1, Rows-1]. For a 3 by 3 kernel, the coordinates of the geometric centre would be [1, 1]. This specification allows the kernel to be skewed with respect to its geometric centre.