For the overwhelming majority of E-O imaging systems, spatial resolution is strictly limited by the detector. This paper studies the main reasons for the influence of the detector on resolution and proposes an imaging method based on special-shaped detectors to enhance the resolution. It is implemented by two detector-arrays of special-shaped pixel where each pixel is reduced by 25% in size. The multiple mis-registered frames are used to solve the grayscale matrix and reconstruct the super-resolution image in the post-detection processing. It presents a theoretical assessment that if a small part of the active area of each pixel is deducted, the frequency-response distribution of the complement will be equal to that of the deducted part independent of the amplitude. To demonstrate the arrangement, a MWIR experiment is undertaken,then two 46 μm×46 μm linear arrays with a 100% full factor are chosen initially and a quarter each pixel is removed. The result demonstrates that the bar-contrast interval is discriminated in an objective space by up to 3.1 mm using an optical system composed of a 6 000 mm f/10 collimator and a 50 mm IR lens. By contrast, it is equivalent to an 11.04 mm with conventional imaging method and 6.9 mm through oversampling technology. The study verifies that the most visually pleasing image or some specific information can be captured by system-matching from optics to electronics for better processing, and including some pre-detection and post-detection processing.