A mathematical approach to describe phase contrast imaging is presented by means of Taylor expansion. The contrast of image of phase object is discussed by changing the absorption and phase-retarding of spatial filters at Fourier plane. A new filtering method is proposed to enhance the image contrast. Based on this approach, a phase-to-intensity image of phase object with higher contrast is got at its imaging plane. Through the image the phase can be detected directly without interferogram processes or phase unwrapping. From the mathematical approach, simulation in MATLAB programming is executed to prove the feasibility of small phase detection. The quick detection of phase may be used for applications such as studying living cells, high throughput microscopy or automated microscopy.