This paper presents a novel image restoration algorithm using examples and truncated constrained least squares (TCLS) filter for ultra-high definition (UHD) television systems. The proposed approach consists of three steps: (i) generation of the patch dictionary using multiple-step image blurring, (ii) selection of the optimum patch based on the orientation and the amount of blurring, and (iii) combination of the selected patch in the dictionary and its filtered version by the TCLS restoration filter for reducing the patch mismatch error. In the proposed algorithm, a complicated point-spread-function (PSF) estimation process is replaced with the generation of multiple, differently blurred patches. Furthermore, the patch dictionary is made by orientation-based classification to reduce the time to search the optimum patch. Experimental results show that the proposed algorithm can restore more natural images with less synthetic artifacts than existing methods. The proposed method provides a significantly improved restoration performance over existing methods in the sense of both subjective and objective measures including peak-to-peak signal-to-noise ratio (PSNR) and structural similarity measure (SSIM).