Lensfree on-chip digital holographic microscope (LFOCDHM) is a simple and efficient miniaturized digital microscope with unique advantages of compact structure, large field-of-view (FOV), and high numerical aperture (NA), which is promising for Internet-of-Medical-Things applications. LFOCDHM imaging generally includes two processes: hologram generation and image reconstruction. In common applications that image stained cells or blood smear samples with relatively high contrast, the threshold segmentation algorithm can be applied to find an object support region (OSR) in the reconstruction process. However, for label-free and low contrast samples, the threshold segmentation algorithm will be unreliable in finding an optimized OSR, which then seriously affects the quality of the reconstruction process. To tackle this problem, the effects of OSR on image reconstruction were studied and an optimized OSR determination method for LFOCDHM was proposed in this work. The results showed that the proposed method can have a 50% lower variance in intensity profiles of reconstructed images compared with traditional threshold segmentation methods.