Automatic image classification by color analysis

An automatic natural scene images classifier and enhancer is presented. It works mainly by combined chromatic and positional criterions in order to classify and enhance portraits and landscapes natural scenes images. Various image processing applications can easily take advantage from the proposed solution, e.g. automatic drive camera settings for the optimization of the exposure, focus, or shutter speed parameters. A large database of high quality images has been used to design and fine tune the algorithm, according to a wide accepted assumption that few chromatic classes on natural images have the most perceptive impact on the human visual system. These are essentially: skin, vegetation and sky-sea. The adaptive color enhancement technique presented, has been designed over the results of the image classifier and it is capable to shift the chromaticity of the regions of interest towards the statistically expected ones, without producing relevant color artifacts. Quantitative results obtained over an extended data set not belonging to the training set, show the effectiveness of the solution proposed both for the natural image classification and the color enhancement techniques.