An Independent Component Analysis Approach for Enhancing Camera Based Low Quality Text Images

The paper proposes an independent component analysis approach specifically aimed at enhancing low-resolution text images from handheld devices. Three layered Independent components are extracted from the image, checked for the components if missing in any of the layer and then integrated so as the missing component can be mixed. Afterwards the image enhancement techniques such as threshold and dilation are used to produce the final text image. Comparative performance evaluation is presented in the form of OCR results of the super-resolution output.