EasyImage - Pre-Processing Images

EasyImage operations prepare images so that further processing gets better results by:

□ Isolating defects using thresholding or intensity transformations
□ Compensating perspective effects (for non-flat surfaces such as a bottle label)
□ Processing complex or disconnected shapes using flexible masks

The main functions are:

□ Intensity transformations change the gray-level of each pixel to clarify objects (histogram stretching Redistribution of the gray-level values of an image, in order to exploit better the available dynamic range, and improve contrast.).
□ Thresholding transforms a binary image into a bi- or tri-level grayscale image by classifying the pixel values.
□ Arithmetic and logic functions manipulate pixels in two images, or one image and a constant.
□ Non-linear filtering functions use non-linear combinations of neighboring pixels (using a kernel) to highlight a shape, or to remove noise.
□ Geometric transforms move selected pixels to realign, resize, rotate and warp.
□ Noise reduction and estimation functions ensure that noise is not unacceptably enhanced by other operations (thresholding, high-pass filtering).
□ Scalar gradient generates a gradient direction or gradient magnitude map from a gray-level image.
□ Vector operations extract 1-dimensional data from an image into a vector, for example to detect scratches or outlines, or to clarify images.
□ Canny edge detector returns a BW8 image of the edges found in a BW8 image.
□ Harris corner detector returns a vector of points of interest in a BW8 image.
□ Overlay overlays an image on top of a color image.
□ Operations on interlaced video frames eliminate interlaced image artifacts by rebuilding or re-aligning fields.
□ Flexible Masks help process irregular shapes in EasyImage.
□ Computing image statistics displays different statistics about an object In a general content, the term object should be understood with the meaning of a class instance. In EasyObject, an object is a maximally-sized area of adjacent connected pixels belonging to the layer foreground. or an image.
□ Fourier transform represents an image by its frequential components to process the image and perform specific filtering.
□ Gabor filter analyzes textures or creates features for a classifier by convolving the image with a wavelet to detect specific frequency content.
□ HDR fusion fuses different images into a single one that contains all the details exacerbated at each exposure level.
□ Image stitching combines partial images to build a final image that appears to have been captured in a single shot.