A Study On Perceptually Inspired Contrast Enhancement Of Images

The main medium of human transmission of information is language and image.In all kinds of information received by man,visual information accounts for 80%.Therefore,the digital image processing technology is very important,which has been developed for half a century,is widely used in various fields of human life and social production,such as aerospace,biomedicine,industrial production and public security etc.Image enhancement is one of the key technologies to improve image display.When the display of the image is too poor,or in order to highlight some of the characteristics of the image,the image enhancement technology is needed to improve the visual effect of the image.Image enhancement technology,according to the given application,to adjust the whole or local information of the image with a purpose,selectively highlight the features of interest in an image or suppress unnecessary features to make the image clear or emphasize certain features.Many scholars at home and abroad in the field of image processing,proposed a number of image enhancement algorithms,has made outstanding contributions for human visual experience.In this paper,we combine a variety of excellent algorithms of the predecessors,use or improve,and put forward our own image enhancement algorithm.In this thesis,the algorithms are combined with the human visual model,so that the results are consistent with the characteristics of the human eye.The main work of this thesis is as follows:1.A threshold histogram equalization algorithm based on human visual model is proposed.Histogram equalization algorithm is a simple and effective image enhancement algorithm,but its shortcomings are obvious.Because it only takes account of the number of gray level pixels,that often lead to over enhancement.Although a lot of people put forward the histogram equalization based on threshold to solve the problem,but it doesn't take into account the human visual system,without considering where to add the threshold value,and how much should be set.The proposed algorithm combined with the human eye model,the appropriate threshold add at the appropriate gray level.2.An low light image enhancement algorithm based on human visual model is proposed.A recent image enhancement algorithm named AGCWD,with a very good enhancement effect.It is mainly characterized by adaptive adjustment factor gamma to enhance the image.However,it sometimes results in a worse effect when dealing with low light images.Especially when the number of pixels with high brightness is small,the algorithm can lead to high gray level being over compressed.According to the principle of Retinex algorithm,the proposed algorithm firstly deals with the illumination layer,generates a light attenuation map,and finally enhances the image by AGCWD.3.An adaptive image enhancement algorithm based on human visual model in DCT domain is proposed.First of all,according to the sensitivity of the human eye to the brightness,the adaptive adjustment factor is used to adjust the AC coefficient.After the detail section is enhanced,the brightness of the whole image is enhanced by CES method which could control the local contrast.