Research On Denoising For Grayscale Image And Passive Autofocusing For Thermal Images

Image noise elimination is an important part in digital image processing.We usually use wiener filtering or linear smoothing filtering to reduce image noise,but the processing effect is not ideal;At the same time,in addition to exploring image denoising from the perspective of image processing,the noise implantation factors can be eliminated in the image acquisition link,thus reducing the image noise generation.Among them,usually poor camera focus is an important factor to produce image noise.Therefore,improving camera focus effect will also help to reduce image noise.Therefore,this paper studies the image noise cancellation technology from two aspects:image noise reduction filtering,and camera automatic focusing technology.The main contents of this paper and the contributions achieved are as follows.1.Integrate the characteristics of noise energy distribution and wavelet transform,a method of image denoising reconstruction based on multi-scale wavelet transform is proposed:first,the noise image is decomposed by high-scale wavelet transform;then the square edge and threshold of white gaussian noise(WGN)or additive white gaussian noise(AWGN)in high-frequency coefficients of wavelet transform are extracted by adjusting the scale of wavelet transform;finally,all threshold coefficients are reconstructed by wavelet transform.The experimental results show that this method can effectively remove the noise in the image and maintain the maximum value of the image information.2.Combining Total Variation Filtering(TVF)and Weighted Bilateral Filtering(WBF),We propose a hybrid Total Variation Filter-Weigh ted Bilateral Filter(TVF-RBF)multi-filter denoising algorithm;also,combining Robust Bilateral Filter(RBF)and Total Variation Filter(TVF)denoising technology,We propose a hybrid Robust bilateral Filter-Total Variation Filter(RBF-TVF)multi-filter denoising algorithm.The proposed hybrid multi-filter method can obtain better image denoising results than single-filter in term of peak signal-to-noise ratio(PSNR)also quantitative comparison experiments.3.For grayscale image denoising,this paper proposes a Bayesian Patch-based image denoising algorithm based on quaternion wavelet transform(QWT):using Patch model instead of Gibbs distributed based energy model.Comparisons include:Expected Patch Log-likelihood(EPLL),Block Matching and 3D Filtering(BM3D),Patch-Based Locally Optimal Wiener(PLOW),Weighted Nuclear Norm Minimization(WNNM),Hybrid Robust Bilateral Filter-Total Variation Filter(RBF-TVF)and Hybrid Total Variation Filter-Weighted Bilateral Filter(TVF-WBF)methods.The experimental results show that the results presented in this paper Bayesian Patch-based image denoising algorithm effectively reduces the noise and the denoising effect is clearer.4.Aiming at the problem of automatic focusing of infrared thermal imaging camera to reduce image noise,this paper proposes a vision and control based auto focus system(VCAFS),which includes:(1)an uncooled thermal camera with motorized lens,(2)a passive contrast-based focus measure,(3)a smoothing operator to avoid local extrema,and(4)two different lens motion controllers.Specifically:an automatic focusing image quality quantization method is further proposed by comparing different focusing methods,a focusing measurement method suitable for thermal imaging is found.In this paper,an automatic focusing thermal image control framework with real-time video feed is proposed to adapt to target scene and depth change.Designed and implemented two configurations of bang-bang controller based on fixed and adaptive step size,and a moving average filter to avoid local extrema,used to control the electric lens of uncooled thermal camera.The experimental results show that the proposed auto-focusing system shows excellent image acquisition quality even in the case of changing scene and depth video.The research work of this thesis enriches the traditional image noise reduction processing theory and technology.