Automatic periodic noise removal in microscopy images

Wavelet and Fourier Decomposition Based Periodic Noise Removal in Microscopy Images

“…For comparison, the methods proposed by F. Sur and M. Grédiac [12] (ARPENOS), J. Varghese [15] (Varghese Filter) and our previous method [16] (FTT9) are used for the same images. To compare the results the Peak-to-peak Signal to Noise Ratio (PSNR) is used.…”

“…Each DWT sub-band is filtered using the method presented in [16]. The magnitude spectrum M is filtered with a sliding window equal with a quarter of minimum value between the height and width of the image.…”

“…In our previous work, [16], we proposed an automated method that analyzes the frequency domain to detect potential noise components in order to remove them. More precisely, the amplitude spectrum is processed with two filters and an iterative thresholding method in order to detect the noise components.…”

“…More precisely, the image is decomposed with 2-D discrete wavelet transform in four bands, each sub-band is processed using the method from [16] and a first filtered image is obtained by recomposing it from the filtered DWT bands. The frequency spectra of the initial image and the one after the first step are used to design two masks for the elimination of periodic noise components.…”

“…The frequency spectra of the initial image and the one after the first step are used to design two masks for the elimination of periodic noise components. Since the methods proposed in [12], [15], [16] have good results, they will be used for comparisons.…”