![]() |
![]() |
![]() |
VIPS Reference Manual | ![]() |
---|---|---|---|---|
Top | Description | Object Hierarchy | Properties |
#include <vips/vips.h> VipsArithmetic; VipsArithmeticClass; void (*VipsArithmeticProcessFn) (struct _VipsArithmetic *arithmetic
,VipsPel *out
,VipsPel **in
,int width
); void vips_arithmetic_compile (VipsArithmeticClass *klass
); VipsVector * vips_arithmetic_get_program (VipsArithmeticClass *klass
,VipsBandFormat fmt
); VipsVector * vips_arithmetic_get_vector (VipsArithmeticClass *klass
,VipsBandFormat fmt
); void vips_arithmetic_set_format_table (VipsArithmeticClass *klass
,const VipsBandFormat *format_table
);
GObject +----VipsObject +----VipsOperation +----VipsArithmetic +----VipsBinary +----VipsUnary
These operations perform pixel arithmetic, that is, they perform an arithmetic operation, such as addition, on every pixel in an image or a pair of images. All (except in a few cases noted below) will work with images of any type or any mixture of types, of any size and of any number of bands.
For binary operations, if the number of bands differs, one of the images must have one band. In this case, an n-band image is formed from the one-band image by joining n copies of the one-band image together, and then the two n-band images are operated upon.
In the same way, for operations that take an array constant, such as
im_remainder_vec()
, you can mix single-element arrays or single-band
images freely.
Arithmetic operations try to preserve precision by increasing the number of
bits in the output image when necessary. Generally, this follows the ANSI C
conventions for type promotion, so multiplying two
IM_BANDFMT_UCHAR
images together, for example, produces a
IM_BANDFMT_USHORT
image, and taking the im_costra()
of a
IM_BANDFMT_USHORT
image produces IM_BANDFMT_FLOAT
image.
For binary arithmetic operations, type promotion occurs in two stages. First, the two input images are cast up to the smallest common format, that is, the type with the smallest range that can represent the full range of both inputs. This conversion can be represented as a table:
Table 1. Smallest common format
in2 /in1
|
uchar | char | ushort | short | uint | int | float | double | complex | double complex |
---|---|---|---|---|---|---|---|---|---|---|
uchar | ushort | short | ushort | short | uint | int | float | double | complex | double complex |
char | short | short | short | short | int | int | float | double | complex | double complex |
ushort | ushort | short | ushort | short | uint | int | float | double | complex | double complex |
short | short | short | short | short | int | int | float | double | complex | double complex |
uint | uint | int | uint | int | uint | int | float | double | complex | double complex |
int | int | int | int | int | int | int | float | double | complex | double complex |
float | float | float | float | float | float | float | float | double | complex | double complex |
double | double | double | double | double | double | double | double | double | double complex | double complex |
complex | complex | complex | complex | complex | complex | complex | complex | double complex | complex | double complex |
double complex | double complex | double complex | double complex | double complex | double complex | double complex | double complex | double complex | double complex | double complex |
In the second stage, the operation is performed between the two identical types to form the output. The details vary between operations, but generally the principle is that the output type should be large enough to represent the whole rage of possible values, except that int never becomes float.
typedef struct { VipsOperationClass parent_class; /* For each input format, what output format. Used for arithmetic * too, since we cast inputs to match. */ const VipsBandFormat *format_table; /* A vector program for each input type. */ VipsVector *vectors[VIPS_FORMAT_LAST]; /* ... and if we've set a program for this format. */ gboolean vector_program[VIPS_FORMAT_LAST]; /* The buffer processor. */ VipsArithmeticProcessFn process_line; } VipsArithmeticClass;
void (*VipsArithmeticProcessFn) (struct _VipsArithmetic *arithmetic
,VipsPel *out
,VipsPel **in
,int width
);
VipsVector * vips_arithmetic_get_program (VipsArithmeticClass *klass
,VipsBandFormat fmt
);
VipsVector * vips_arithmetic_get_vector (VipsArithmeticClass *klass
,VipsBandFormat fmt
);
void vips_arithmetic_set_format_table (VipsArithmeticClass *klass
,const VipsBandFormat *format_table
);