Interface Description

The interfaces of the CV image operator are the same as those of OpenCV. Table 1 lists optimized interfaces. To facilitate adaptation, Interface Definition uses OpenCV function interfaces as examples.

**Table 1** Optimized interfaces
Interface	Description
cvtColor	Converts images, including the BGR2GRAY, BGR2RGB, RGB2BGR, and GRAY2BGR modes.
dilate	Dilates the target area in an image by convolving the image based on specified structure elements.
GaussianBlur	Uses the Gaussian function to perform convolution on images to smooth images and reduce noise and details.
merge	Merges three single-channel images into one three-channel image.
remap	Remaps images during image processing for geometric transforms, such as image translation, rotation, scaling, and distortion.
resize	Adjusts the image size in INTER_NEAREST_EXACT (nearest neighbor interpolation), INTER_LINEAR (bilinear interpolation, default), and INTER_CUBIC (bicubic interpolation) modes.
convertTo	Converts the data type of an image to another with optional scaling. For example, an image can be converted from 8-bit to 16-bit and scaled down at the same time.
threshold	Performs image binarization based on a threshold that divides the grayscale value of an image into two parts. Methods such as global thresholding, adaptive thresholding, and OTSU thresholding can be used.
warpAffine	Performs affine transformation such as translation, rotation, scaling, and cropping on an image using a 2x3 transformation matrix.
contourArea	Calculates the area of a contour, that is, the total number of pixels in the contour.
countNonZero	Counts the number of non-zero pixels in an image.
morphologyEx	Performs morphological operations, such as opening, closing, and gradient operations. It is used to process noise, fill holes, or enhance edge features in an image.
copyMakeBorder	Adds a border around an image. It expands the image size by filling a specific color or replicating border values to enhance the image processing effect.
bitwise_and	Performs a bitwise AND operation to combine the pixel values of two images and retains only pixel information that exists in both input images.
bitwise_or	Performs a bitwise OR operation to generate a composite image that includes at least one input image having non-zero pixels.
kcv_normalAndNCHW	Performs normalization, the result of which is used as the input of deep learning models.
t	Performs matrix transposition to swap the rows and columns of an image matrix to meet specific computing requirements.

The usage of the interfaces and involved constants are the same as those in OpenCV. To avoid tedium, this document only briefly describes some interfaces. For more information, refer to the official OpenCV documentation.

Parent topic: CV Image Operators