FrameWork for separable filters
dip_Error dip_SeparableFrameWork (in, out, boundary, border, process )
The dip_SeparableFrameWork function is a framework for separable filters. This function takes care of all the "administrative work" involved when processing a n-D DIPlib image n times with a 1-D filter. In short, using this function, one has only to create an one dimension filter function and dip_SeparableFrameWork takes care of the other stuff. The in image is filtered nrOfProcesses times using the information in each element of the process array. If nrOfProcesses is zero, only the first element of process is used to filter in in all its dimensions.
| Data type | Name | Description |
| dip_Image | in | Input image |
| dip_Image | out | Output image |
| dip_BoundaryArray | boundary | Boundary conditions |
| dip_IntegerArray | border | Border array |
| dip_FrameWorkProcessArray | process | Array of dip_FrameWorkProcess structures |
The dip_FrameWorkProcess structure contains the following members:
| Data type | Name | Description |
| dip_Boolean | process | flags specifying to do processing or not |
| dipf_FrameWorkMethod | frameWorkMethod | flags specifying the method of how dip_SeparableFrameWork should transport data from in to out |
| dipf_FrameWorkOperation | frameWorkOperation | flags specifying requirements of the 1-D filter function |
| dip_int | processDimension | dimension of in to be processed |
| dip_int | roiOrigin | ignored in current implementation |
| dip_int | roiSize | ignored in current implementation |
| dipf_FrameWorkFilter | FrameWorkFilterType | specifying the type of 1-D filter function |
| dip_FrameWorkFilter | FrameWorkFilter | pointer to the 1-D filter function |
| void * | functionParameters | parameters of the 1-D filter function, for all dimensions |
| dip_DataType | inputBufferType | data type of input buffer |
| dip_DataType | outputBufferType | data type of output buffer |
| dip_DataType | suggestedOutputType | data type of output image |
The dipf_FrameWorkMethod enum contains the following elements:
| Name | Description |
| DIP_FRAMEWORK_DEFAULT_METHOD | use dip_SeparableFrameWorks most optimal method |
| DIP_FRAMEWORK_CLASSICAL | use a classical method |
| DIP_FRAMEWORK_DOUBLE_STRIPE | use two buffers to store temporary results |
It is strongly advised to use the DIP_FRAMEWORK_DEFAULT_METHOD method.
The dipf_FrameWorkOperation enum contains the following elements:
| Name | Description |
| DIP_FRAMEWORK_DEFAULT_OPERATION | default operation |
| DIP_FRAMEWORK_IN_PLACE | filtering operation can be performed in-place. It is up to dip_SeparableFrameWork whether the actual filtering is done in-place |
| DIP_FRAMEWORK_NO_IN_BORDER | the 1-D filter does not need border extension of the input data |
| DIP_FRAMEWORK_OUT_BORDER | the 1-D filter needs border extension of the output data |
| DIP_FRAMEWORK_WRITE_INPUT | the 1-D filter needs to write in the input data |
| DIP_FRAMEWORK_USE_BUFFER_TYPES | made the input and output buffers of the inputBufferType and outputBufferType data type |
| DIP_FRAMEWORK_NO_BUFFER_STRIDE | Create input and output buffers with a stride of one |
| DIP_FRAMEWORK_DO_NOT_ADJUST | Do not adjust the output image, just check it |
| DIP_FRAMEWORK_USE_OUTPUT_TYPE | Adjust output image to the suggestedOutputType data type |
The dipf_FrameWorkFilter enum contains the following elements:
| Name | Description |
| DIP_FRAMEWORK_SEPARABLE_FILTER | default filter type, process one line at the time |
| DIP_FRAMEWORK_TWO_LINES_SEPARABLE_FILTER | process two lines in one go |
| DIP_FRAMEWORK_SINGLE_OUTPUT_FILTER | this filter only needs an output buffer |
The union dip_FrameWorkFunction consists of the types
| Name | Description |
| dip_SeparableFilter | one line filter function |
| dip_TwoLinesSeparableFilter | two lines filter function |
| dip_SingleOutputFilter | single output image filter |
The functions have the following arguments: dip_SeparableFilter
| Data type | Name | Description |
| void * | inData | pointer to the input data |
| void * | outData | pointer to the output data |
| dip_int | elements | number of pixels in the inData array |
| dip_SeparableFilterParameters | params | parameter structure for the filter function |
and dip_SeparableTwoLinesFilter
| Data type | Name | Description |
| void * | inFirstLineData | pointer to the data of the first input line |
| void * | inSecondLineData | pointer to the data of the second input line |
| void * | outFirstLineData | pointer to the data of the first output line |
| void * | outSecondLineData | pointer to the data of the second output line |
| dip_int | elements | number of pixels in the inFirstLineData array |
| dip_TwoLinesSeparableFilterParameters | params | parameter structure for the two lines filter function |
The inData, inFirstLineData and inSecondLineData will always point to the first pixel of the line of in that is processed. Therefore, the 1-D filter can access pixels with indices ranging from -border[dimension] up to elements + border[dimension]. If the flag DIP_FRAMEWORK_OUT_BOUNDARY is specified, the same holds for the output data pointers.
The structure dip_SeparableFilterParameters contains the following elements:
| Data type | Name | Description |
| void * | functionParameters | parameters of the 1-D filter function per dimension |
| dip_int | dimension | the dimension in which direction the input buffer is taken from the input image |
| dip_int | processNumber | number of times dip_SeparableFrameWork has already filtered in with an 1-D filter including current filtering |
| dip_DataType | inType | dip_DataType of the input buffer |
| dip_int | inStride | stride of the elements in the input array |
| dip_int | inPlane | plane number in case in is a binary image |
| dip_DataType | outType | dip_DataType of the output buffer |
| dip_int | outStride | stride of the elements in the output array |
| dip_int | outPlane | plane number in case out is a binary image |
| dip_int | outDimension | the dimension in which direction the output buffer is taken from the output image |
| dip_IntegerArray | position | coordinate of the first pixel of the input buffer in the input image |
The structure dip_TwoLinesSeparableFilterParameters contains the following elements:
| Data type | Name | Description |
| void * | functionParameters | parameters of the 1-D filter function per dimension |
| dip_int | dimension | the dimension in which direction the input buffer is taken from the input image |
| dip_int | processNumber | number of times dip_SeparableFrameWork has already filtered in with an 1-D filter including current filtering |
| dip_DataType | inType | dip_DataType of the input buffer |
| dip_int | inStride | stride of the elements in the input array |
| dip_int | inPlane | plane number in case in is a binary image |
| dip_DataType | outType | dip_DataType of the output buffer |
| dip_int | outStride | stride of the elements in the output array |
| dip_int | outPlane | plane number in case out is a binary image |
| dip_int | outDimension | the dimension in which direction the output buffer is taken from the output image |
| dip_IntegerArray | position | coordinate of the first pixel of the input buffer in the input image |
The structure dip_SingleOutputFilterParameters contains the following elements:
| Data type | Name | Description |
| void * | functionParameters | parameters of the 1-D filter function per dimension |
| dip_int | dimension | the dimension in which direction the input buffer is taken from the input image |
| dip_int | processNumber | number of times dip_SeparableFrameWork has already filtered in with an 1-D filter including current filtering |
| dip_DataType | type | dip_DataType of the input buffer |
| dip_int | stride | stride of the elements in the input array |
| dip_int | plane | plane number in case in is a binary image |
| dip_IntegerArray | position | coordinate of the first pixel of the input buffer in the input image |
DIPlib's data types SeparableConvolution, MonadicFrameWork, SingleOutputFrameWork, PixelTableFrameWork, ScanFrameWork