Theories And Methods For Intelligent Image Enhancement And Image Quality Evaluation Based On Fidelity Criteria

The image is the result of the interaction between the radiation and the object.In the case of optical surface imaging,the image represents the distribution of the reflection coefficient of the object.In terms of optical surface digital imaging,the digital image represents the reflection coefficient matrix of the object.The scene condition is poor and the visual effect is not good,it is not conducive to the observation and subsequent processing.So the image enhancement techniques are needed to improve the visual quality of the image.The study found that the image with the higher the contrast is not always the better the image visual quality.In the sense,image enhancement should be accurately referred to as the visual effect enhancement of image,rather than a simple contrast enhancement.The essence of image enhancement is to sacrifice some secondary information in exchange for the improvement of image visual quality.Modern image science is divided into the theory and technology of image acquisition and image processing.Image processing can be divided into storage,transmission and cognition.Image compression services to storage and transmission,image enhancement services to image recognition.All ofthese technologies and processes need to be high fidelity and ideal without distortion,such as lossless compression.A good or desired technique and method is the least distortion one.This also applies to image enhancement.Image enhancement is science,not art.It should be based on facts,not on psychological pleasure.On the basis of analysis and test existing image enhancement methods may produce distortion,the study points out the key scientific problems is fidelity in the field of existing image enhancement,put forward four fidelity criteria in image enhancement,and suggested the theory and method on measurement the fidelity of the enhanced image.Based on the universal theory of thermodynamics,this paper proves that the image enhancement is the process of "entropy reduction" which makes the image quality artificially better,which provides the theoretical basis for the establishment of the four criteria of the image enhancement.This paper puts forward that "the essence of image enhancement is the process of sacrificing some secondary information in exchange for the improvement of image visual quality".This definition of image enhancement implies "entropy reduction",which is to sacrifice some secondary information rather than increase information.The image quality assessment in the field of image compression is studied and the essence of the assessment in this field is to assess the authenticity.In this study,the quality evaluating indicator of image enhancement and image enhancement method based on fidelity are also established.It is proposed and demonstrated that image brightness is the core element of image quality parameters,and according to experiments,it is proved that there is a corresponding optimization image for each scene.The focus of this study is to study the distortion or fidelity of image enhancement which is seldom discussed at present.On the basis of the theory of fidelity in image enhancement,the theory and method of measurement fidelity/distortion are put forward.Based on the analysis of some existing image enhancement methods with the fidelity of defects,we present several new and intelligent image enhancement methods which conforms to the fidelity principle and enhanced image is quality optimization,mainly including Zadeh-X transformation,intelligent bi-directional preserving bandwidth logarithmic transformation and intelligent luminance transformation.Experiments also show that the methods are superior to the current mainstream methods such as CLAHE and MSRCR,which many parameters usually need to be set when using these methods,but several methods of this study do not need to set parameters,and they are completely new adaptive intelligent image enhancement methods.