Automatic focusing method based on image processing

Abstract

The invention discloses an automatic focusing method based on image processing. The definition evaluation function values of every n position points are used as a section of local curve for 'fitting'in the hill climbing search process; fine search is realized by judging whether the curve contains a slope peak greater than a set threshold value or not. The method comprises the following steps: dividing a small range with a wave crest into n points again, reducing the step length, repeatedly searching, continuing to move n position points along the original direction if a slope crest which is greater than a set threshold does not exist in a local curve, and repeating the operation to find the optimal focusing position. According to the method disclosed in the solution, an improved Robert gray function is used as an image definition evaluation function; the thought of a curve fitting method is embedded into the local search of the traditional hill climbing algorithm; the method not onlyinherits the simple reliability of a traditional hill climbing algorithm, but also can effectively get rid of the interference of a pseudo-focus peak; the method prevents the problem of complexity andtime consumption caused by falling into a local peak value in the focusing search process, and can be widely applied to various imaging systems such as cameras and microscopes.

Description

technical field [0001] The invention belongs to the technical field of image processing and automatic focusing, and in particular relates to an automatic focusing method based on image processing. Background technique [0002] With the rapid development of intelligence and automation of various imaging equipment, autofocus technology has become one of the key technologies in various types of imaging systems and computer vision. Autofocus refers to the process of obtaining a clear image on an image detector (such as a CMOS or CCD detector) by adjusting the position of the lens during image acquisition. According to its focusing principle, automatic focusing can be divided into two types: active focusing and passive focusing. Active focusing is mainly based on distance measurement. By measuring the distance between the target and the lens, the accurate focusing position is calculated accordingly, and the moving device is directly driven to reach the position to complete the f...

AI Technical Summary

Problems solved by technology

[0006] However, due to the variety of target scenes and different lighting conditions, it is difficult for the sharpness evaluation function curve to be a monotonous and smooth single-peak curve in real situations, but to present multiple pseudo-focus peaks and local Oscillation (such as figure 1 Shown is the comparison between the sharpness evaluation function curve and the actual curve in the ideal state), which makes the focus search very easy to fall into the local peak, resulting in too long focusing time or even focusing failure

In response to this problem, many scholars have proposed improvement schemes, such as curve fitting search algorithm, Fibonacci search algorithm, slope-based adaptive step-size search algorithm, etc., but these algorithms have low stability, poor adaptability or resistance. Many problems such as poor noise performance, it is difficult to popularize and apply

For example, based on the slope-adaptive step-size search algorithm, the slope cannot be defined effectively only by relying on the sharpness evaluation values ​​of the two points before and after, so there is no strong correspondence between the slope and the step-size at the local peak, and the search failure rate is still low. High; the effect of the curve fitting search algorithm is better near the extreme value, but it is very dependent on the data near the extreme value. Once a large error occurs in individual data, it will have a great impact on the overall fitting result, resulting in Offset of extreme points

Images