The star tracker estimates pointing knowledge of a satellite in arcsecond accuracy in three axes without apriori knowledge. But star trackers are larger in size, heavier, power hungry and expensive for nanosatellite missions. The Arcsecond Pico Star Tracker (APST) is designed based on the limitations of nanosatellites and estimated to provide pointing knowledge in arcsecond. The APST is developed using fully COTS components because it's affordable, and less development time. A theoretical model is developed to estimate the performance of the COTS components (image sensor, lens, and baffle) used in the APST. Using this model, it's possible to validate if the components meet the requirements of the star tracker. But COTS component decreases the overall performance due to the errors in image sensor noise, lens distortion, and aberration etc. In APST, the lens distortion and inaccurate centroiding are the dominant error sources. The radial lens distortion causes an error in angular distance measurement, which leads misidentifying or identification of stars and high processing time. This leads to functional failure of APST. To overcome this, the relative star selection method is iv developed which selects the stars based on the angular distance information. Based on the fact that star pair with low angular distance has minimum measurement error, the relative star selection selects the four stars with low measurement error. It's compared with conventional bright star selection method, whereas stars are selected based on brightness. The relative selection algorithm is tested with 75-star constellation in star simulator and it has delivered 100% success rate and accuracy of 71 arcseconds in boresight. Whereas the conventional bright star selection delivered low success rate of 28% because the star pairs are not selected based on angular distance separation. Hence the relative star selection algorithm is efficient for APST.