Histogram Equalization Algorithm Based On Otsu Controllable Grayscale Stretching

Images are an important piece of information that humans perceive the world.Therefore,image information is also the most direct and intuitive information that human beings encounter in daily life,production and consumption,and scientific research.Its application field must involve all aspects of human production,life and work.However,due to the shooting environment of the picture,transmission interference,storage error,human error,and other reasons,the recognition of the picture is reduced,which causes certain difficulties for the later processing and application.The result is that the quality of an image is often unsatisfactory.However,the application of median filtering,average smoothing,homomorphic filtering and histogram equalization makes the image effect enhanced and improves the appearance of the image,enhancing the details of the lowbrightness image,making the grayscale range of the enhanced image wider.The contrast has been improved.Among them,global quadrature equalization is a widely used contrast enhancement method.However,because it acts on the whole picture,the parts of the picture are stretched without difference,and the detail is enhanced,which inevitably leads to the amplification of noise and the change of picture brightness.As a result,the picture is distorted as a whole,and even the picture quality is degraded.In order to overcome the above shortcomings,this paper proposes a method of partitioned quadrature equalization,which uses the maximum inter-class variance method to first partition the original image,and then adopts different degrees of stretch for the feature information of different partitions in the partitioned image.In the subsequent experiments,it is proved that the partitioned quadrature equalization algorithm proposed in this paper effectively overcomes the local over-enhancement and under-enhancement effects and the effect of noise amplification in the global enhancement algorithm.This allows the original image to maintain its original realism while achieving an overall contrast enhancement.In the subsequent experiments,it was found that in the face image engineering dealing with uneven illumination,since the difference between the shaded part and the highlighted part is too large,the highlight and shadow parts will be placed after the maximum inter-class difference areais used.Among the different partitions,the proposed algorithm further compresses the grayscale range of the shaded area and enlarges the grayscale range of the highlighted area,thereby causing the difference between the two parts to further expand in the enhanced image.Such enhanced pictures have certain difficulties in the later processing or use.Especially in the current face recognition algorithm is even more unsatisfactory.In order to solve such a problem,a gamma correction algorithm is added before the partitioned quadrature equalization,so that the difference between the shaded portion and the highlighted portion of the original picture is balanced,and the enhanced picture can bring a good visual experience to the viewer.In order to overcome such a phenomenon,this paper combines gamma correction to make the enhanced picture more conducive to human visual characteristics.The results show that the algorithm is superior to other related algorithms in terms of detail and visual realism by partition enhancement.