Abstract: A novel wedge-ring poor-pixels photoelectric detector is valuable for a micro vision system. However an image acquired by the detector has extremely low resolution and it does not reflect the same or similar shape information of an object in the real world. To enable such a detector and its images available in further object identification, a unique shape recovery framework was presented in this paper. By rotating the wedge-ring detector around its center in a sub-wedge range, original low-resolution images were generated. Then linear interpolation along with a least squares method was applied to preliminarily recover the object shape. After noise removal via a two-stage level set evolution with an edge indicator function, the final high-quality object shape was achieved. Experiments demonstrate the effective performance of the proposed algorithms, in which the shape recovery rate is up to 95%.