基于图像识别技术的摄像机自动聚焦系统设计

A Design of the Camera's Auto-focusing System Based on the Technique of Image-recognition

国内外众多关于摄像机的自动聚焦算法都存在响应的实时性差、系统稳定性不够,特别是在低照度下拍摄时,由于噪声的增加,还容易出现误焦的情况.在基于离散余弦变换的修正频谱评价函数基础上,结合传统摄像构图采用的黄金分割理论,选取了5个关键点作为聚焦参考点并进行处理,以提高系统的聚焦响应灵敏度;同时为了提高系统的抗干扰能力和稳定性,采用对采集到的数字图像进行预处理的方式,以减少噪声对聚焦动作的干扰.实验表明,本系统在低照度环境及高照度环境下都能够获得比较清晰的图像,有较高对比度,被拍物的细节特征丰富明显,具有灵敏度高、聚焦可靠性强等特点,可广泛应用于各类摄像机中.

Nowadays,in auto-focusing algorithm there exits poor real-time quality and systematic unstability,home or abroad.Especially there is even the wrong focus under low brightness because of noise.Based on the theory of discrete cosine transform and the golden-section method applied in traditional compositional visual-recording,this paper illustrates how to set 5 reference focus points to improve the system sensitivity of the camera and how to use the collected digital images to increase the anti-interference capability and stability of the camera system.The experiment reveals that such new system can obtain relatively clearer images under both low and high illumination with a sharp contrast.Furthermore the characteristics of the taken object are rich and obvious.Because of its high sensitivity and strong focusing ability,this new system can be widely applied to various cameras.