Enhancement Of Low Illuminated Flower Image Based On Retinex

Flower images collected in low illuminated environment show the properties of low overall brightness and insignificant contrast between flowers and background,which do not meet the perceptual standards of human visual system and practical application requirements.Therefore,preprocessing of low illumination flower images is the key to improving the visual quality.In this paper,researches on flower images collected under low illuminated conditions are carried out,and image enhancement algorithms for low illumination flower images are designed.The main work of this paper is as follows:(1)In order to realize low illuminated flower image enhancement,a variational optimization model based on Retinex illumination theory was improved.Firstly,this paper proposes three kinds of prior information,including illumination intensity prior,structure consistency prior and edge perception prior.The prior information of illumination was constructed by combining the pixel intensity of RGB three channels,and the prior information of structural consistency was constructed according to the linear mapping relationship of pixels in the local area before and after image enhancement,and the coefficients of the linear mapping model were regularized.The prior information of image edge perception is constructed ensure that the algorithm could smooth the high-frequency details and preserve the low-frequency structure information more completely,and add it to the model as a weighting function.Then,the optimization model was established by combining the three kinds of prior information,and the method of alternating iteration minimization was applied to solve the model.Finally,a variational optimization image enhancement algorithm based on the joint prior is proposed.After simulation experiments,qualitative analysis of the enhanced image of the improved algorithm,and quantitative evaluation using LOE and VIF index show that the improved algorithm in this paper is better than other improved algorithms,reflecting the effectiveness of the algorithm.(2)The optimization model is provided in logarithmic domain to make the enhancement effect of low illumination flower image more consistent with human visual perception.Firstly,an adaptive weighting function was constructed for the illumination and reflectance components,respectively,and the corresponding regularization terms are constrained to achieve enhancement of low illuminated flower image while preserving more structure and detail information.Then,combining the illumination prior information constructed by the three channels of the color image,a regularization term is established.Finally,the optimization model was established by combining the illumination prior regularization term and the adaptive weighting function,and the alternative minimization method was used to solve the model,and the weighted optimization image enhancement algorithm based on the illumination prior was constructed.The simulation experiment shows that the NIQE index of the image enhanced by the improved algorithm is lower than that of the comparison algorithm,which reflects better naturalness and effectiveness of the algorithm.