Correction of Optical Flow Calculations Using Color Balance Change

To detect the detailed motions of moving objects, we propose a new correction method for optical flow calculations based on the block matching algorithm. After first analyzing optical flow results obtained using the block matching algorithm, the three features were found. Based on these three features, we constructed a correction method for optical flow calculations. In the developed correction method, optical flow images are multiplied. Some optical flow images were calculated from an image for which the color balance was changed, while others were calculated from neighboring frames. At the multiplication stage, the vector length of the optical flows were gathered based on their frequency distribution. To evaluate the proposed method, experiments were conducted using video images in which a color checker is moving to the left or right. It is found that several erroneous optical flows are calculated in the case of the before correction, while the number of errors is drastically reduced in the case of the after correction.