The innate character of ink jet instability sometimes causes device defects in polymer light emitting diodes (PLED). This study develops an image-processing method to identify defects by methods such as filtering, color processing, background subtraction, dilation, etc. Based upon the defect information and the pre-built ink jet print head library, the amount of ink volume and the location to print on can be appropriately estimated to patch defects. In addition, an optimal printing route can be determined with a specially designed head driving waveform, consisting of pre-oscillations during motion intermission. By this methodology, a 60 pL sized drop from ink jet nozzles can approximately patch the PLED panel with pixel resolution of 240×73 μm, the QCIF standard. The results demonstrated good accuracy in landing while patching unfilled pixels. However, there are some satellite drop occurrences observed due to the sensitivity of image data threshold level setting. Overall, the confidence level of successful defect identification is larger than 95%. In this paper, an optimal repair route and a strategy about thin film morphology control were also discussed. These approaches are helpful to repair defects and, consequently, increase the yield rate for PLED or other applications by ink jet technology.