Abstarct: Technical limitations in image capturing device may impose defective in captured images. Blurriness is one of the imposing defectives. Blurriness is created via elimination of high frequency components (edges and details) in image, which leads to contrast degradation. There are various approaches to improve the contrast of images. In most of them, prior knowledge about the original image is required to enhance low contrast images. In practice, this knowledge is not available with a good accuracy. Besides, image enhancement process without any prior knowledge is usually time consuming and a complicated task. Among all these approaches, un-sharp masking method has attracted more attention because of its simplicity in implementation and calculation, and without requirement to a prior knowledge. Un-sharp masking techniques enhance image contrast via boosting high frequency components of the image. In the classic un-sharp masking technique, at first, the high frequency components of the input image is extracted via using a linear high pass filter; then by adding a scaled amount (the gain factor) of these components to the input image, a higher contrast image is obtained. The gain factor has an important influence on the un-sharp masking results. In classic un-sharp masking, a constant value is considered for this parameter, without assessing the image content and the amount of blurriness. Hence, it is an important problem to choose an appropriate value for the gain factor. Besides, the amount of blurriness in the image may be different in each region of the image. Hence, it is necessary to determine different gain values for each region of the image. In this thesis, the gain factor is determined adaptively considering the image content and the amount of blurriness in different regions of the image. To achieve this goal, a global and local approaches are proposed in this thesis. Gradient information of an image can provide information about the contrast of the image. Hence, in global approach, a gain factor is determined for the whole input image via assessing the image gradient information. Blocking or segmentation methods can be used for local blur image enhancement. For local image enhancement via blocking method, the input image is divided into non-overlapping blocks. Then, the desirable gain value is determined for each block via assessing the gradient information of the block. In local image enhancement via segmentation, the input image is segmented into blur and non-blur regions baxsed on the amount of blurriness. Then, the desirable gain value is determined for each segment via assessing the gradient information of the segment. The CSIQ, Berkeley, and VOC2007 databaxses were used to compare performance of the proposed methods with other existing methods. Experimental results show the superiority of the proposed un-sharp masking method compared to the existing methods in image enhancing.