ABSTRACT Human recognition based on face images has a challenging task for the variations in image intensities and pose angles. This paper proposes an efficient face recognition system using compressed hybrid domain features. The benchmarked face image databases were resized to 112 X 92. The Viola Johnes algorithm is used for the face part detection and resized to 80 X80, then converted image pixel decimal values into 8-bit binary. The binary bits are segmented into Most Significant Bits (MSB) and Least Significant Bits (LSB) and converted each 4-bit binary into corresponding decimal values, then reshaped into the matrix of size 80 X 80. The Discrete Wavelet Transform (DWT) is applied on MSB, whereas the Histogram of Gradients (HOG) is used on LSB matrices. The GIST concept is used on the LL-sub band of DWT to extract the first set of initial features and HOG features as the second set of initial features. The final compelling features are attained using the convolution of GIST and HOG features. The Artificial Neural Network (ANN) is used to categorize the features of face image databases and test images to verify the system’s performance. The accuracy of the anticipated procedure is better to the existing approaches.