Fourier Analysis and Image Processing Homework Help

Images can also be represented through its frequency. Animage is a multi-color graph unless the image is completely black or white. This non-uniformity in picture incorporates the concept of frequency in image processing. A significant difference between pixels results in a large frequency while a uniform picture will have a lesser magnitude of frequency. The frequency related content of an image can be obtained using Fourier Analysis[4]. To avoid complexity involved in applying a multi-dimensional transformation, images are broken into vertical and horizontal pixels, transformation is then performed separately.A computer can easily identify edges using frequency because an edge has a higher frequency as compared to a surface. Simulink has a separate block for frequency analysis through which an image is easily transformed into a discrete form. Conversion of image from real-object to digital form may inculcate certain amount of noise that is filtered through frequency based filters. Noise may be acceptable for few applications but for crucial ones, even a little amount is unacceptable. For medical purposes, that little inaccuracy can be harmful. The filters eliminate or accept a certain range of frequencies, the cutoff frequency is specified by the user. We do not apply the transform on every frequency multiple of image but only to those which have high weightage in the formation of image. It is observed that low frequency pixels possesses more image content as compared to high frequency ones.

Simulink provides features to transform imagery data into frequency components. The same syntax is applicable for big data with minor changes, hence making Simulink suitable for big datasets as well. A machine differentiates images through comparison and there is a technique of image signals for comparison. Convolution evaluates the similarity and correlation between two signals. Convolution also helps in determining the amount of filtering required by comparing different pictures. Computer performs convolution between previously stored image data and currently read image for detection.

Image filtering is categorized into two classes; Pre and Post Processing filtering. First filtering technique filters the image before compression, however, the latter one applies transforms after compression. Pre-processing is performed by predicting the intensities based on the characteristics of stored images. This techniques avoids a significant amount of computational delay. Instead of filtering a picture as a whole, algorithms divide the image into two vectors separately containing different dimensions.

Share This