The various low level features are extracted based on their visual content which are color, shape, texture etc.

  1. Color Feature Extraction Techniques

    1. Color Histogram defines the probability of intensities of the three color channels. Histogram of image is not changed to the rotation and translation of the image plane, and it is changed slowly when the angle view is changed

    2. Color Correlogram gives the information about how the colors pairs are changed with distance

      A color correlogram of an image is a table obtained by computing the number of color pixels of j at a distance k from color i gives the kth entry at location (i,j), divided by total number of pixels in the images.

      1. it describes the correlation of colors in spatial plane
      2. it is used to describe the global distribution by local spatial correlation of colors
      3. it is relatively simple in computation
      4. The feature size is equitably small

    3. Dominant Color Descriptor describes the typical colors in an image or image region. It is used for retrieving similar images from database and browsing of image database based on single or various color values.

    4. Color Co-occurence Matrix capturing color variations in the image which gives the color feature. calculate the probability of the occurrence of same pixel color between each pixel and its adjacent pixel

      Untitled

  2. Texture Feature Extraction Techniques visible patterns which have likeness properties that do not possible from existence of only a single color or intensity

    1. Tamura Texture Feature six textual features: coarseness, contrast, directionality, line-likeness, regularity and roughness (First three features are generally most important!)

    2. Steerable Pyramid divides an image into a set of oriented sub-bands and low-pass residual. The image is decomposed into a set of undecimated directional sub-bands and one decimated lowpass sub-bands

    3. Wavelet Transform texture analysis and classification using multi-resolution approach

      Untitled

    4. Gabor Wavelet Transform most widely used technique for texture analysis feature extraction.

    Untitled

    Untitled

reference

https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7566544