Color Correlation Assessment And Intelligent Color Correlation Algorithm

Along with the widespread use of the image color correction algorithm in the image processing field,an increasing number of color correction algorithms have been presented.Meanwhile,corresponding assessment metrics remain insufficient because of lacing of color distortion database.To achieve objective color correction image quality assessment(IQA)that are consistent with subjective perception,we create an Image Color Correction Database(ICCD).ICCD contains 6 types of distortions in brightness,hue,saturation,exposure,contrast,and R and G channels.Each type has three different scales.We select six state-of-the-art color correction algorithms to perform color correction for each target image.Then we design and conduct user study to get users' Mean Opinion Score(MOS).We assess color correction results from the perceptive of IQA.Most of existing IQA methods are not designed for assessing color distortion,and require reference and target images capture the same scene,which cannot be applied directly to assess images with similar scene.To solve the problem,we use SIFT Flow(Scale-Invariant Feature Transform Flow)algorithm to achieve regional registration between images and generate a matching image.Besides,we introduce confidence map of matching image and saliency map of target image to improve the accuracy.The experimental results show that our proposed metric has better correlation,accuracy,and monotonicity than all compared state-of-the-art metrics.Furthermore,we propose a machine learning based color correction assessment to mine and synthesis the effects of numerous factors.The features used for learning are from our image-registration-based IQA methods and image resizing quality assessment.The experimental results show that our machine learning based assessment metric synthesizes the advantages of each single assessment metric and further improves the performance.Finally,based on our second proposed color correction assessment metric,we propose a parameter automatic choosing algorithm and a correction results adaptive fusion algorithm for color correction quality enhancement.The parameter automatic choosing algorithm can select the best values for color correction models so as to enhance the quality of correction results.The fusion algorithm considers the quality of each correction result,and generates a better fusion result.