Billions of documents to download for free
A low-contrast image enhancement method based on the Xiaodu transform is proposed. The multi-scale characteristics of wavelet transform are used. Different filtering strategies are adopted for the energy, detail and noise of the image to enhance the image as a whole. The method is implemented by programming in VC ++ environment, and the experimental results show that the enhanced algorithm can obtain images with good overall visual effects.
Weipu Information http://bardagim.com
A pair of jealous and slow honey aj pairs] Degree spoon == I = Wu Ding Huang Chaobing, Jiang Yingying (Information of Chinese University of Science and Technology: School of Engineering, Wu Si Wuhan 4 0 7) 3 0 0 Abstract: A low-contrast image enhancement based on Xiaodu transform is proposed method. Multi-scale characteristics of wavelet transform are used.
And noise sections adopt different filtering strategies. Enhance the image as a whole. This method is implemented programmatically in the v + environment. C +. Experimental results show that the enhanced algorithm can obtain images with good overall visual effects. Keywords: image contrast enhancement; wavelet transform; multi-degree analysis
Introduction Image enhancement is a very important research area in image processing. At present, there are many mature and effective methods1 such as: I histogram, equalization, pass filtering, high inverse mask sharpening, etc., but these methods exist The problem is that the noise is amplified. Wavelet analysis is a new time-frequency analysis tool developed in recent years. It has time-frequency localization and multi-resolution analysis capabilities. It is particularly suitable for signal processing and image processing. There have been many research results on the application of wavelet transform in image enhancement. Some simply apply wavelet transform methods to image enhancement. As the image is enhanced, it also amplifies the noise. The "threshold" and "hard threshold" wavelet image enhancement methods using soft or "can effectively suppress noise while enhancing the image [￣, but sometimes ringing occurs. This paper proposes a low-contrast image enhancement method based on 2t-wavelet transform, which adopts different filtering strategies for the possible amount, detail, and noise of the image. The overall visual effect of the obtained image is good.
The band contains more vertical high-frequency information. Correspondingly, in the L l band, there are mainly high-frequency components in the horizontal direction, and the frequency band H HH is the reflection of high-frequency information in the diagonal direction of the image, especially at a height of 4 j 5 degrees or 15 degrees Frequency information. 3 ld} HL 2 LL, H 【b HH l + H H k H Hz
Figure 1 Schematic diagram of image wavelet transform
2 Contrast-enhanced image After wavelet decomposition, it can be changed by linear or non-linear
Method to enhance the relationship of images in different frequency bands at different resolution scales
1 The wavelet transform of the image adopts a two-dimensional wavelet transform with a bi-orthogonal property to the image. It is achieved by the one-dimensional wavelet transform of rows and columns, respectively, using 1.
The wavelet coefficients of an image are obtained. Use from re
Inverse transformation can be obtained
To contrast-enhanced images. The specific implementation method is as follows:
Generation 1t speed algorithm. A fast NN image is decomposed by a layer of two-dimensional Mal fast algorithm x lt a. The wavelet decomposition of N2 × N is calculated, and four subband images with the size of (/) (/ machine 2 are obtained. After the wavelet decomposition of the i layer.
① Load image data. Complete the three wavelet transforms to obtain the wavelet coefficients w; xy (= ,,,) j 12 3 represents the number of wavelet decomposition layers, = 1 (i0 ,, 2 3 represents the corresponding L, H, L ,, LLH HH band) ( ,) Pixels, y is x
The position coordinates of the point in the image.
The schematic is shown in Figure 1. All
② Wavelet coefficient regularization processing, because the wavelet coefficient of the wavelet transform may exceed 2 5 or even more. 5 For this purpose, the range of the wavelet coefficient needs to be mapped into the N [, 5] o 2 5 interval.
Each level of wavelet decomposition of the image data always divides the upper-level low-frequency data into finer frequency bands. The middle HL band is first put on it;
Four-level low-frequency image data is f-direction after low-pass filtering in the horizontal direction),
Nine through high-pass filtering (direction) in the vertical direction. Therefore, the HL frequency period is listed
③ The filter function gx = K— () ix with the smallest calculation amount is selected to realize the filtering processing of wavelet coefficients.
@ MD NC PE OE OURRMT