In the measurement of particle size by Fraunhofer diffraction method based on a linear CCD, the Chin-Shifrin inversion algorithm can lead to false peaks in the inversion of Particle Size Distribution (PSD). To overcome this phenomenon of the algorithm, a rectangular window function was proposed and introduced in this algorithm. The midpoint and boundary of the window function were determined by analyzing the relationship between particle size and its minimum value of derivative of the diffraction light intensity. The inverted PSD was truncated by superposing the window function to remove the false peaks and enhance the accuracy of the inverted PSD. The results of the inverted PSDs obtained by using different algorithms were compared by measuring two types of standard materials respectively. Experimental results show that, the improved Chin-Shifrin algorithm can effectively eliminate the false peak distributions in the inverted PSD. The relative error of the measuring results is less than 3%, and the repeatability is no more than 4%.