A novel approach for content based image retrieval in context of supervised learning and regression analysis

A growing number of researchers try to find image query equivalents of the successful text-pattern matching and text-indexing techniques. In this proposed work, a generalization of the text-retrieval methodologies is hindered by the aforementioned characteristics of multimedia image. Several stages are included for measuring proper residuals. The distance of pixels is measured in HSV format to ensure the better accuracy than RGB format. More over R-Square indication provides the better regression line that fits the data. Significance F and P values are also calculated for recognizing the variable's statistical significant. It deletes a variable with a high P-value (>0.05) and rerun the regression until significance F drops below 0.05.