Research On Image Enhancement Based On Adaptive Threshold

With the development of the era of science and the progress of technology, people's expectation on image quality is higher than before. However in real life, some images have significant noise or their texture is vague. Some images have poor visual effect. The effect of some classical enhancement methods cannot meet the requirements of people. Therefore, new methods should be created to improve the ability of enhancement on the basis of the original method. This paper briefly introduces the development and application of image enhancement technology, and the image enhancement in the defogging and medical application. This paper introduces the evaluation standard of image enhancement. This paper mainly does the following work:Firstly, it introduces several common enhancement method. Spatial domain method is that the image of the basic unit pixel processing, through the division of neighborhood, median, mean, Wiener, Fourier transform and mathematical means of image pixels are recalculated as new image pixels. Image enhancing formation is reconstructed by the new image pixels. Wavelet uses the multiresolution ability. The images are devided by the frequency domain, mainly through threshold segmentation method. The wavelet method enhances classification coefficient of low frequency and high frequency attenuation decomposition coefficients. It highlights the outline part, weakening noise and reconstructing the denoised image. In recent years, block matching algorithm is on the rise in denoising algorithms and here we introduces the three dimensional block matching algorithm based on image block similarity and the application of the joint filtering of each block.Secondly, according to the results of previous studies, this paper proposes the adaptive weighted vector filtering method based on image texture feature adaptive neighborhood selection and set the center pixel as a benchmark. The feature pixels which is different from the center pixel in the neighborhood is stored in a vector. Then we reconstruct the image after the extraction of the median and maximum. The experimental results prove its superiority. By the calculation, the improvement of PSNR to the median method is an average increase of 5.259dB, to the wiener method is an average increase of 10.326dB and to the BM3D method is an average increase of 1.9737dB.Thirdly, because of the poor effect of histogram adjustment and wavelet for the fuzzy enhancement method and image sharpening, we proposes a new algorithm of adaptive diagonal symmetry. The proposed adaptive average pixel threshold processes pixels on the basis for classification. It improves operator by using symmetry principle and enhances the pertinence and effectiveness of the algorithm. Experimental results show that the modified method has improvements and progress.Image enhancement is an important part of image processing and lays a solid foundation for further processing and research to improve the visual effect of the image. Image enhancement has a broad application prospect.