Image enhancement is the process of making an image more interpretable for a particular application (Faust, 1989). Enhancement makes important features of raw, remotely sensed data more interpretable to the human eye. Enhancement techniques are often used instead of classification techniques for feature extraction—studying and locating areas and objects on the ground and deriving useful information from images.

The techniques to be used in image enhancement depend upon:

- Your data—the different bands of Landsat, SPOT, and other imaging sensors are selected to detect certain features. You must know the parameters of the bands being used before performing any enhancement. (See Raster Data for more details.)
- Your objective—for example, sharpening an image to identify features that can be used for training samples requires a different set of enhancement techniques than reducing the number of bands in the study. You must have a clear idea of the final product desired before enhancement is performed.
- Your expectations—what you think you are going to find.
- Your background—your experience in performing enhancement.

Enhancement techniques available with ERDAS IMAGINE are as follows:

- Correcting Data Anomalies - radiometric and geometric correction
- Radiometric Enhancement - enhancing images based on the values of individual pixels
- Spatial Enhancement - enhancing images based on the values of individual and neighboring pixels
- Resolution Merge and Wavelet Resolution Merge - fusing information from several sensors into one composite image
- Spectral Enhancement - enhancing images by transforming the values of each pixel on a multiband basis
- Hyperspectral image processing - This section is superceded by the Spectral Analysis User Guide
- Independent Component Analysis - a high order feature extraction technique that exploits the statistical characteristics of multispectral and hyperspectral imagery
- Fourier Analysis - techniques for eliminating periodic noise in imagery
- Radar Imagery Enhancement - techniques specifically designed for enhancing radar imagery

See Bibliography to find literature that provides a more detailed discussion of image processing enhancement techniques.

Display vs. File Enhancement

With ERDAS IMAGINE, image enhancement may be performed:

- temporarily, upon the image that is displayed in the Viewer (by manipulating the function and display memories), or
- permanently, upon the image data in the data file.

Enhancing a displayed image is much faster than enhancing an image on disk. If one is looking for certain visual effects, it may be beneficial to perform some trial and error enhancement techniques on the display. Then, when the desired results are obtained, the values that are stored in the display device memory can be used to make the same changes to the data file.

Reference: For more information about displayed images and display device memory, see Image Display.

Spatial Modeling Enhancements

Two types of models for enhancement can be created in ERDAS IMAGINE:

- Graphical models—use Model Maker (Spatial Modeler) to easily, and with great flexibility, construct models that can be used to enhance the data.
- Script models—for even greater flexibility, use the Spatial Modeler Language (SML) to construct models in script form. SML enables you to write scripts which can be written, edited, and run from the Spatial Modeler component or directly from the command line. You can edit models created with Model Maker using SML or Model Maker.

Although a graphical model and a script model look different, they produce the same results when applied.

Image Interpretation

ERDAS IMAGINE supplies many algorithms constructed as models, which are ready to be applied with user-input parameters at the touch of a button. These graphical models, created with Model Maker, are listed as menu functions in the Image Interpretation functions. Just remember, these are modeling functions which can be edited and adapted as needed with Spatial Modeler or Model Maker.

See Geographic Information Systems for more information on Raster Modeling.

The modeling functions available for enhancement in Image Interpretation functions are briefly described in the following table.

Function | Description |

SPATIAL ENHANCEMENT | Enhance the image by using the values of individual and surrounding pixels. |

Convolution | Uses a matrix to average small sets of pixels across an image. |

Non-directional Edge | Averages the results from two orthogonal 1st derivative edge detectors. |

Focal Analysis | Perform one of several analyses on class values in an image file using a process similar to convolution filtering. |

Texture | Defines texture as a quantitative characteristic in an image. |

Adaptive Filter | Varies the contrast stretch for each pixel depending upon the DN values in the surrounding moving window. |

Statistical Filter | Produces the pixel output DN by averaging pixels within a moving window that fall within a statistically defined range. |

Resolution Merge | Merges imagery of differing spatial resolutions. |

Crisp | Sharpens the overall scene luminance without distorting the thematic content of the image. |

RADIOMETRIC ENHANCEMENT | Enhance the image by using the values of individual pixels within each band. |

LUT (Lookup Table) Stretch | Creates an output image that contains the data values as modified by a lookup table. |

Histogram Equalization | Redistributes pixel values with a nonlinear contrast stretch so that there are approximately the same number of pixels with each value within a range. |

Histogram Match | Mathematically determines a lookup table that converts the histogram of one image to resemble the histogram of another. |

Brightness Inversion | Allows both linear and nonlinear reversal of the image intensity range. |

Haze Reduction* | Dehazes Landsat 4 and 5 TM data and panchromatic data. |

Noise Reduction* | Removes noise using an adaptive filter. |

Destripe TM Data | Removes striping from a raw TM4 or TM5 data file. |

SPECTRAL ENHANCEMENT | Enhance the image by transforming the values of each pixel on a multiband basis. |

Principal Components | Compresses redundant data values into fewer bands, which are often more interpretable than the source data. |

Inverse Principal Components | Performs an inverse principal components analysis. |

Decorrelation Stretch | Applies a contrast stretch to the principal components of an image. |

Tasseled Cap | Rotates the data structure axes to optimize data viewing for vegetation studies. |

RGB to IHS | Transforms red, green, blue values to intensity, hue, saturation values. |

IHS to RGB | Transforms intensity, hue, saturation values to red, green, blue values. |

Indices | Performs band ratios that are commonly used in mineral and vegetation studies. |

Natural Color | Simulates natural color for TM data. |

FOURIER ANALYSIS | Enhance the image by applying a Fourier Transform to the data. |

Fourier Transform* | Use a highly efficient version of the Discrete Fourier Transform (DFT). |

Fourier Transform Editor* | Edit Fourier images using many interactive tools and filters. |

Inverse Fourier Transform* | Computes the inverse two-dimensional Fast Fourier Transform (FFT) of the spectrum stored. |

Fourier Magnitude* | Converts the Fourier Transform image into the more familiar Fourier Magnitude image. |

Periodic Noise Removal* | Automatically removes striping and other periodic noise from images. |

Homomorphic Filter* | Enhances imagery using an illumination and reflectance model. |

* Indicates functions that are not graphical models.

There are other Image Interpretation functions that do not necessarily apply to image enhancement.