In order to overcome the puzzle that in the complex illumination condition, the traditional LBP image feature recognition algorithm neglects the contrast between local areas and abandons some important texture feature information, which makes it difficult to eliminate the nonlinear distortion caused by intense illumination changes, this paper proposed an improved image feature recognition algorithm. The algorithm first normalizes the face image, and then projects the pixel contrast value into a certain interval with the help of the ALBP algorithm, due to that the image after processing has good illumination invariance, the image features can be recognized better by using this method. The results of comparative experiment demonstrated that the recognition rate of ALBP algorithm is 84%, and that of LBP algorithm is 75%, the former is 9% higher than the latter. The research shows that the ALBP recognition method can eliminate the nonlinear distortion of the image well under complex illumination.
|