The influence of the sampling pinholes and recording aperture on imaging quality in coherent diffractive imaging based on sampling array is analyzed using scalar diffraction theory. A mathematical formula for describing the influence is derived. The theoretical analysis shows that the limited size of the recording aperture can result in a crosstalk between adjacent sampling pinholes. This crosstalk effect is the main source of the wavefront reconstruction error and the imaging noise. Through digital analysis of the changes of the sub maximum of the crosstalk function under different recording aperture conditions, it is found that the sub maximum does not descend monotonously when the size of the recording aperture increases, while it oscillatorily decreases with enlargement of the recording aperture. An optimal size of the recording aperture is defined and an analytical formula is given. Further experimental results show that the background noise will descend to its minimum when the radius of the recording aperture is just equal to the optimal size.