摘要
在油气资源勘探开发领域，超声成像测井以其图像直观、全井眼覆盖和探测范围大等优点，得到了广泛应用。超声成像测井不但可以在裸眼井中反映井眼几何形状，识别裂缝、孔洞、层理等地层非均质性，而且还能在套管井中检查射孔质量、分析套管损坏以及评价固井质量。但是由于超声成像测井过程中复杂的测井作业环境以及超声换能器非理想性声斑等因素，导致了超声测井图像模糊，造成了图像的对比度降低，分辨率下降，为细小目标地质体特征的分析和解释带来了困难。
在此背景下，本文结合超声成像测井的工程需求，以超声成像测井原理和信息处理为理论基础，采用物理模拟和数值模拟相结合的方法，从信号产生与处理的角度开展了超声测井图像增强和复原方法研究。本文主要工作如下：
1.基于限定对比度直方图均衡的超声测井图像增强方法的研究
针对超声测井图像增强问题，基于超声成像对比度低的客观实际，结合工程实时性的要求，研究并实现了HE，BBHE，RMSHE，POSHE，BOHE，MLBOHE，CLAHE等多种直方图均衡方法。在对算法性能分析的基础上引入幂次变换方法，通过参数调整对灰度级进行非线性变换。分析了子块数量、剪切阈值、幂次等参数的选择与增强效果的关系。采用幂次变换与CLAHE方法相结合，提出了CLAHE-PL图像增强方法。利用实验室超声图像、模型标准井和油田现场测得的超声测井图像进行了对比验证测试，主
观评价和客观指标（MG、PSNR、AMBE、IE和LC）评价均指示了CLAHE-PL方法增强超声测井图像是有效的。
2.基于部分重叠的直方图均衡超声测井图像增强方法的研究
小米鸡蛋粥>对联特点针对部分重叠的直方图均衡方法存在过度增强的问题，将CLAHE算法中剪切直方图引入到POSHE方法中，修改子块累积直方图分布函数。以低对比度的夜间车辆监控图像为例，分析了剪切阈值的大小对增强效果的影响，提出了POSHEOC 图像增强方法，利用平均梯度和平均结构相似度指标构建测井图像质量评价的策略，实现了最佳的子块直方图剪切阈值的自动选取。在此基础上，以模型标准井和2幅油田现场实测的超声测井图像为例，结合PMGSIM、PSNR、IE、AMBE 和LC等5个客观评价指标，将本文提出的POSHEOC和CLAHE-PL两种算法与HE、BBHE、RMSHE、POSHE 、BOHE和MLBOHE等6种方法进行对比测试。实验
结果表明，POSHEOC方法处理超声测井图像是有效的，实现了提高超声测井图像对比度的同时，限制了平坦区过度增强，突出了局部细节信息。
3.基于APEX点扩展函数估计的最佳K值维纳滤波复原方法的研究
针对超声测井图像复原问题，分析了图像退化的因素，明确了声束的扩散在井壁形成的声斑是造成超
声测井图像退化的主要原因。基于超声测井成像系统退化模型未知的实际情况，研究了声波换能器的传输特性，利用试验估计法在退化图像中选取特征点作为参考对象，推导出点扩散函数模型近似为G类函数。采用APEX算法，对模型参数进行估计，将APEX参数拟合拓展到水平和垂直两个方向，进行对数幅度谱截面曲线拟合，估计出点扩散函数。基于估计的点扩散函数，利用维纳滤波方法进行图像复原，针对维纳滤波复原方法中噪信比K值未知的问题，提出了基于引导滤波和最小均方误差自动找寻最佳噪信比K值的方法。在此基础上，基于已知的标准清晰图像，利用计算机模拟退化图像，并增加不同功率的高斯噪声，开展复原性能测试，并结合MG、PSNR、IE、AMBE 和LC等5个客观评价指标，对复原图像的质量进行了对比分析，验证了所提复原算法的正确性。在模拟退化图像进行复原正确的基础上，利用实验室超声图像、模型井超声图像和现场测井图像进行了大量的仿真实验，结合主观观察和客观评价指标，开展复原图像的性能测试。实验结果表明，本文提出的基于APEX点扩展函数估计的最佳K值维纳滤波复原方法对超声测井图像的复原是有效的，在恢复图像细节信息，增强清晰度，抑制噪声方面均表现出良好的性能。
郭作贵本文的主要成果：
1.研究了基于直方图的系列图像增强方法，提出适合超声测井图像增强的两种方法CLAHE-PL和POSHEOC，所提方法可以内嵌到现场测井资料处理与解释软件中，服务于油气勘探开发，亦可推广适用于水下声学图像的增强。
2.研究了基于试验估计的图像盲复原方法，并估计了超声图像的退化模型，将APEX算法引入到超声测井图像点扩散函数的参数估计中，提出了基于引导滤波和最小均方误差自动找寻最佳噪信比K值的方法，探索出了一种超声测井图像退化模型估计方法和测井图像复原算法，研究成果在油田勘探开发中有较好的应用前景。
关键词：超声成像测井，图像增强，图像复原，直方图均衡，点扩散函数
Abstract
Due to the advantages of covering the entire borehole wall with intuitive image and large detection range, ultrasonic well logging has been widely ud in the field of exploration and development of oil and gas resources.ultrasonic imaging logging can not only reflects the borehole geometry in a borehole, identify formation heterogeneity that includes fractures, holes and texture, but also determines the quality of perforation ,analyzes the casing damage and evaluate cementing quality in cad hole well. However, during the process of ultrasonic imaging logging, impacted by the complex logging operating environment and the non-ideal sound spot of the ultrasonic transducer , ultrasonic logging images often suffer from low contrast and indistinct local details, which makes it difficult to analyze and interpret geologic features of small target in the images.
Under this background, according to the engineering requirements of ultrasonic imaging logging , bad on the principles of ultrasonic imaging logging and information processing , from the perspective of signal generation and processing , this disrtation us a combination of physical simulation and numerical simulation to do some rearch on the method of image enhancement and restoration of ultrasonic logging . The main work of this disrtation is as follows:
1.Rearch on the method of enhancement method for ultrasonic logging image bad on contrast limited adaptive histogram equalization
For the problem of ultrasonic logging image enhancement, bad on the drawback of low contrast for ultrasonic image, according to the requirements of real-time in engineering, many histogram equalization methods including HE, BBHE, RMSHE, POSHE, BOHE, MLBOHE, CLAHE were rearched and implemented .Bad on the algorithm performance analysis, the power-law transformation method is introduced to perform non-linearly transformation on the gray level through parameter adjustment. The relationship between the enhancement effects and lection of parameters such as the number of sub-blocks, cut-off threshold, power is analyzed. Combined with the
苏州莲花岛
正与邪
CLAHE method and power-law transformation, the CLAHE-PL image enhancement method is propod. Comparison and verification tests were conducted on the ultrasonic images of laboratory equipment，model well and field logging to test the performance of the propod method.Experimental results bad on visual perceptual evaluation and quantitative measures(MG, PSNR, AMBE, IE, and LC) showed that the propod method CLAHE-PL is effective in enhancing the ultrasonic logging image.
2.Rearch on the method of enhancement method for ultrasonic logging image bad on partially overlapped sub-block histogram-equalization
For the problem of excessive enhancement in the partially overlapped sub-block histogram-equalization, the idea of contrast-limited enhancement is introduced to modify the cumulative distribution functions of the POSHE . Taking the night vehicle surveillance image with low-contrast as an example, the effect of different cut threshold on the enhancement performance is analyzed, and the POSHEOC image enhancement method is propod. This method build a new quality evaluation index considering the effects of the mean gradient and mean structural similarity. The new index is designed to obtain the optimal clip-limit value for histogram equalization of the sub-block. It makes the choice of the optimal clip-limit automatically according to the different input image. taking model 卡通足球
玩积木well and two ultrasonic logging images measured in the field as an example, combined with five objective evaluation indicators such as PMGSIM, PSNR, IE, AMBE and LC, the performance of the POSHEOC and CLAHE-PL propod in this disrtation are compared with other six methods including HE, BBHE, RMSHE, POSHE, BOHE and MLBOHE. Experimental results bad on visual perceptual evaluation and quantitative measures demonstrate that the propod method is effective in processing ultrasonic logging images and yields better quality in enhancing the contrast, emphasizing the local details while prerving the brightness and restricting the excessive enhancement compared with the other histogram equalization bad techniques from the literatures.
3.Rearch on the best K-value wiener filter restoration method bad on APEX
point spread function estimation
元宵节的习俗有哪些Aiming at the problem of ultrasonic logging image restoration, the factors of image degradation are analyzed, and it is clarified that the sound spot formed by the sound beam diffusion on the wall is the main reason for the degradation of ultrasonic logging images.Under the fact that the degradation model of the ultrasonic logging imaging system is unknown, the transmission characteristics of the acoustic wave transducer are studied, and the feature points are lected as reference objects in the
degraded image by the test estimation method, and the point spread function model is derived as class G function. Using the APEX algorithm, the model parameters are estimated, the APEX parameters fitting are extended to the horizontal and vertical directions, the curve fitting of log-amplitude spectrum cross-ction is performed, and the point spread function is estimated. Bad on the estimated point spread function, the Wiener filtering method is ud for image restoration. For the problem of the unknown value of the noi signal ratio K in the Wiener filter restoration method, an automatic arch for the optimal noi signal ratio K bad on guided filtering and minimum mean square error is propod. Bad on the known clear image, computer is ud to simulate the degraded image with the class G function and Gaussian noi of different powers are added to carry out the restoration performance test. Combining five objective evaluation indicators including MG, PSNR, IE, AMBE and LC, the quality of the restored image was analyzed to verify the correctness of the propod restoration algorithm. On the basis of restoration test for simulated degraded images, a large number of simulation experiments were carried out on the ultrasonic images of laboratory equipment，model well and field logging to test the performance of the propod method.Experimental results bad on visual perceptual evaluation and quantitative measures demonstrate that the propod optimal K-value Wiener filter restoration method bad on the APEX point spread function estimation is effective for the restoration of ultrasonic logging images, and it shows good performance in restoring image details, enhancing clarity, and suppressing noi.
The main achievementof this disrtation:
