I would to compress images using machine learning, what kind of preprocessing can use before applying machine learning?
There are many image preprocesing techniques. These are used for noise removal, contrast enhancement and illumination equalization. For noise removal we use various filters such as medican filter, wiener filter, gaussian filter etc. For contrast enhancement and illuimination we can use contrast stretching. Apart from these many other tecniques are available.
Pre-processing is a common name for operations with images at the lowest level of abstraction -- both input and output are intensity images. The aim of pre-processing is an improvement of the image data that suppresses unwanted distortions or enhances some image features important for further processing. Four categories of image pre-processing methods according to the size of the pixel neighborhood that is used for the calculation of a new pixel brightness are: 1. pixel brightness transformations
2. geometric transformations
3. pre-processing methods that use a local neighborhood of the processed pixel, 4. image restoration that requires knowledge about the entire image.
no any definite process or techniques .It is depends on the type and acquisition of image.
The kind of pre-processing depends on type of imaginary system i.e. medical images, remote images because the image degraded during the acquisition process and the image should be restored before compression. You have to quantify degradation function and improve image before compression.
There are a number of pre-processing techniques as suggested by the above peer groups. Best of my knowledge, for an effective and efficient pre-processing of an image in the context of compression, it is better to use "Confidence Limit-based" pre-process at a particular significant level. The significance level could be fixed according to your requirement. By this technique, the outliers could be reduced considerably with a less computational time complexity.
Preprocessing depends on type of images(medical, remote ,etc.), if the image is true color, and other considerations .
No fixed method of pre processing. Methods is depends upon the type of images.
many methods are there Bi lateral filter ,AHE and CAHE gives good results
- Badges
- Science topic