Edge Detection with multiscale products for SAR image despeckling

Summary:

Recording the image that comes from Aparture Synthetic Radar (SAR) many use in various applications, but the image of the recording of many mixed the spots are very difficult to remove. This paper aims to perform edge detection in SAR image recording can effectively
maintain conditions for the removal of the edge image patches of the image. Using multiscale products without the basic information can be build a domain Stationary Bundlet Transform (SBT) to detect the edge of image and SBT is used to eliminate the interference or noise on the edges of the image. algorithm edges detection do first then proceed with building algoritam despeckling.

To calculate the threshold used,

after the parameter set of SBT. Test results when compared with Lee Filtering on conformity
higher where Region I, Lee (20.6166) and MPs SBT (66.1432) with the test other equivalent accuracy. In conclusion, this paper proposes a method Multiscale edge detection using the Product in the Domain Theatre Bundlet Transform (SBT) and the results are used for cleaning detected
patches - patches on SAR images to maintain the edge of the image during the process
removal of the patches thus this method is better than Lee Filtering.

Advantages:

• This method can retain the edge of image for the removal of blemishes.

Disadvantage:

• This method performs two processes that edge detection and the removal of blemishes, that time is needed for longer than using single algorithm on a large scale.

Suggestions:

• There are good algorithms detect and descpeckling edge join to reduce the complexity of the calculations.

Contributions:

• Reduces spots - spots that exist on the recording of the SAR image.The proposed approach is to use a multiscale products during the reduction process can keep the edge of image patches.