Abstract: The auto-focusing process applied to integrated camera is described as the lens looking for the maximum value of image definition evaluation function according to auto-focusing search algorithm. The auto-focusing search algorithm has a great influence on the focusing time and focusing accuracy of the whole auto-focusing process. In order to solve the problem that the existing algorithms fail to focus under light source scenes and no-detail scenes, a focus search algorithm based on scene prediction is proposed.The algorithm can intelligently analyze, judge and classify imaging scenes, and uses different search strategies in different scenarios to improve the stability and accuracy of auto-focus searching.At the same time, aiming at the problem that the existing algorithm is difficult to determine the direction of the peak of focus curve in defocus area, an fuzzy degree evaluation model is introduced, which can accurately determine the peak direction of the focus curve. Analysis of experimental data shows that the new algorithm has increased the focusing speed in the defocus area by 24.3%. Experiments verify that the new algorithm has high application value in the integrated camera imaging system based on Hass Hi3518A processor.