Analytical autofocus technology method
3D Panel Fence
3D Panel Fence,3D Mesh Fence,3D Wire Fence,3D Wire Mesh Fence HeBei Bosen Metal Products Co.,Ltd , https://www.hbbosenfence.com
Business: Seiko lens domestic lens CCD infrared camera
Differentiating from the basic principle of automatic focusing, automatic focusing technology can be divided into three categories:
1. Ranging method This method is to achieve auto focus by measuring the distance between the subject and the lens and driving the lens to the appropriate position. Specific methods include triangulation, infrared ranging and ultrasonic ranging.
2, focus detection method This method is to achieve automatic focus by detecting whether the image is out of focus or focus (mainly detecting the offset of the contour edge of the image and the image). Specific methods include contrast method and phase method.
3, image processing method This method is to capture the optical image of the measured object through the camera, through analog and digital conversion, using the computer for digital image processing to achieve automatic focus. The specific method is divided into a focus distance evaluation function: a high-frequency component method, a smoothing method, a threshold integration method, and a gray-scale difference method.
The various autofocus techniques are briefly described above, but these methods have their limitations. For example, the focusing method of infrared and ultrasonic ranging can only perform detection control at the head (a lens with infrared or ultrasonic waves attached to both sides of the camera lens), which is difficult in the arrangement of the process structure and inconvenient to maintain. In addition, when the target is strongly absorbed by infrared or ultrasonic waves, the system will malfunction and the focus will be inaccurate. However, its advantage is that it is not easily affected by other scenes. In the contrast method of focus detection, it is also subject to the illumination condition. If the illumination condition is weak or the difference between the subject and the background is small, the focus adjustment will be difficult or even lose its effect.
With the development of computer and image processing technology, the automatic focusing method based on image processing has developed rapidly. Because of the flexibility of the computer to process images, various other valid information in the image can be extracted for different usage requirements, thereby selecting different focus criteria for focus determination. Therefore, it is widely used in today's camera and measurement systems. Especially in solid-state cameras and digital cameras, they require fast and accurate focusing.
Generally, whether the image is accurately focused or not is reflected in whether the boundary and the detail part of the image are clear in the spatial domain, and whether the high frequency component of the image is rich in the frequency domain. The edge and detail information of the image can be obtained by differentiating the image; the spectral information of the image can be obtained by performing FFT (ie, fast Fourier transform) on the image. Image differentiation method (ie gray level difference method) Although the algorithm is simple and fast, it can't filter out the noise in the image. The image FFT method (that is, the high-frequency component method) allows the spectrum to pass through the band-pass filter after FFT. High-frequency information filters out higher-frequency noise at the same time, but the algorithm is more complicated and time-consuming. The image processing method can be used to obtain the differential amplitude or spectral amplitude of the image, and the maximum position of the amplitude is the optimal focus position of the image.
From the above analysis, in the current situation, in order to obtain accurate, simple and fast auto-focusing in digital cameras and cameras, image differentiation method can be used together with smoothing processing, or filtering noise before differentiation, generally adopted Five-point median filtering eliminates the second largest peak in the function curve due to noise. In addition, due to the high correlation of the video signal, the number of sampling points should not be too much, and it is better to sample 200 points per frame. This not only reduces the probability of white noise, but also suppresses white noise.
3D Wire Fence Introduction: