To improve the quality of maternity care and the processing speed, an accurate determination of gestational age from ultrasound images is very essential. The biometric parameters at second semester vary in accuracy from 2 to 3 weeks. The manual placement of calibers and probe during measurement also introduces user variability in prediction. In this paper, an attempt has been made to estimate the gestational age at third trimester by biparietal diameter (BPD) and kidney length (KL). The kernel-based fuzzy C means clustering, circular Hough transform (CHT) and generalized fuzzy Hough transform have been used for the determination fetal skull and kidney shape. The use of gradient vector in CHT works only in relevant orientation and reduces the computation time. An active contour model is then imposed on the structure to detect features such as BPD and KL. The pixel-by-pixel scanning technique has been used to extract features of an image which is simulated using MATLAB program, and the measured results were very close to real measurements. The performance of the present work was evaluated with 50 ultrasound B mode images in gestational age range from 28 to 36 weeks; each has different shapes and sizes. The obtained results have been compared with the standard measurements, and it is found that the measurement error is limited toý±ý8.7%. It can also be improved with the help of kidney length using indirect measurements. It reduces the measurement error toý±ý3.08%.