Name
Title | ||
|---|---|---|
An Analysis of Adaptive Filtering and Double Filtering on the Enhancement of Images in the Wavelet Domain |
Abstract | ||
|---|---|---|
The objectives of this paper is to illustrate the enhancement of images in the wavelet domain using double filtering technique and compare the performance with adaptive filtering for Gaussian noise. The two algorithms are also evaluated on the Blackfin DSP processor for their computational overhead in an embedded system. Sample 'CT' scanned and MRI images are used for analysis. Double filtering is achieved using linear estimation and non-linear filtering. Wiener method performs estimation based on the statistics of the local neighborhood pixel on a window of 3x3 blocks. This method is the first pass intending to eliminate any noise, which can be additive or multiplicative. The second pass of the filter is implemented using median filter using a mask of 5x5. The adaptive filter is implemented using the LMS algorithm. The computational overhead of the LMS algorithm in an embedded system is evaluated by fixing a constant step size and the number of iterations for convergence. Simulated study is carried out to obtain an enhanced image from a noisy image. The enhanced image is compared with the original image to determine the performance of double filtering and adaptive filtering. The performance of the technique towards any type of noise can be measured by calculating the peak signal to noise ratio. The performance evaluation of the enhancement algorithms on the processor is evaluated by determining the total number of cycles for a single pixel. |
Year | Venue | Keywords |
|---|---|---|
2007 | Lecture Notes in Engineering and Computer Science | LMS algorithm,median filter,wavelet decomposition,wiener filtering |
Field | DocType | ISSN |
Pattern recognition,Computer science,Filter (signal processing),Artificial intelligence,Adaptive filter,Kernel adaptive filter,Machine learning,Wavelet | Conference | 2078-0958 |
Citations | PageRank | References |
0 | 0.34 | 1 |
Authors | ||
2 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Bob Paul Raj | 1 | 0 | 0.34 |
| V. Ramachandran | 2 | 8 | 2.93 |