CN105072330A - Linear array camera automatic focusing method - Google Patents

Abstract

The invention discloses a linear array camera automatic focusing method, belonging to the image acquisition field, and especially relates to a linear array camera focusing method. The method comprises: first changing the distance between a shot object and a camera, and meanwhile collecting a target image by a linear array camera; analyzing each collected image and the size of values of image pixels, determining an image not to meet requirements if the image is all black or white, and re-collecting images; splicing all images collected in a whole distance change process into an image; performing gray-scale treatment, de-noising treatment and enhancing treatment on the spliced image; and finally calculating the clearest position in the image, wherein the position at which the image of an object corresponding to the clearest position is collected is a best shooting position. Compared with the prior art, the method has the advantages of low cost, fast speed and small computational complexity.

Description

A kind of automatic focusing method of line-scan digital camera

Technical field

Belong to IMAQ field, particularly a kind of focus adjustment method of line-scan digital camera.

Background technology

Present stage requires higher resolution and precision, makes to adopt line-scan digital camera more and more in the large image scanning field of elongated, belt-shaped.In order to ensure certainty of measurement, wherein most critical is exactly the automatic focusing of line-scan digital camera, and it can affect the picture quality collected, and later image process also can be made to meet difficulty, and even cannot carry out.

Traditional machine visual detection device based on line-scan digital camera, need on the basis of line-scan digital camera increase an area array cameras for auto-focusing, this method problems faced have following some:

1. add equipment cost;

2. equipment needing the space additionally increasing an area array cameras, also needing to carry out extra maintenance to area array cameras when debugging, mechanical complications and the debugging complexity of equipment all increase a lot.

3. traditional face battle array focusing principle is when out of focus, mobile object lens, and gather two pictures and carry out definition contrast, then object lens move to the direction that definition is high.This method needs to gather plurality of pictures and carries out definition analysis, and focusing speed is slow and efficiency is low.

Summary of the invention

The present invention is directed to the weak point of background technology, design the automatic focusing method of the line-scan digital camera that a kind of cost is low, speed is fast, amount of calculation is little.

Technical scheme of the present invention is a kind of automatic focusing method of line-scan digital camera, and the method comprises:

Step 1: change the distance between shot object and camera, with line-scan digital camera, target image is gathered simultaneously;

Step 2: analyze each two field picture collected, analyzed by the size of the value to image slices vegetarian refreshments, if image is entirely black or entirely white, then image is undesirable, re-starts collection;

Step 3: it is piece image that whole distance change procedure is obtained all image mosaic;

Step 4: gray processing process, denoising, enhancing process are carried out to stitching image;

Step 5: to calculate in this image position the most clearly, position during the collected image of object corresponding to this position is optimum photographing position.

The concrete steps of wherein said step 5 are:

Step 5.1: little picture segmentation is carried out to stitching image;

Step 5.2: adopt energy gradient function to carry out preliminary definition judgment to the little picture that 5.1 split, finds in picture of publishing picture and splits picture more clearly;

Step 5.3: adopt wavelet transformation to find out the pixel that in the segmentation picture that step 5.1 obtains, definition is the highest.

The advantage of this method has:

The picture collected have employed energy gradient function and comes Primary Location articulation point position, avoids and uses wavelet transformation to cause a large amount of calculating at the very start; Arithmetic speed can be caught up with actual productive temp completely, finally use wavelet transformation accurately to locate and also ensure that accuracy rate, can the automatic focusing of extensive use line-scan digital camera, large degree saved production cost.

Accompanying drawing explanation

Fig. 1 is that the automatic focusing method of a kind of line-scan digital camera of the present invention adopts drawing system schematic diagram;

Fig. 2 is that the automatic focusing with fuzzy-clear-fuzzy gradual change collected gathers image;

Fig. 3 is the articulation curve calculated with gradient function algorithm, abscissa presentation video coordinate, and ordinate represents definition, and value is more large more clear;

Fig. 4 is the articulation curve calculated by Wavelet Transformation Algorithm.

Embodiment

Step 1: position object being placed on distance camera certain distance, and light filling operation is carried out to object.Then reduce the distance between shot object and camera, with line-scan digital camera, target image is gathered simultaneously;

Step 2: each two field picture using multithread analyzing to collect, is analyzed by the size of the value to image slices vegetarian refreshments, if image is entirely black or entirely white, then image is undesirable, re-starts collection;

Step 3: it is piece image that whole distance reduction process is obtained image mosaic;

Step 4: use floating-point arithmetic to carry out gray processing process, carry out denoising based on medium filtering to stitching image, use the linear transformation in greyscale transformation, between the gray area at outstanding interested target place, between those uninterested gray areas of relative suppression, the dynamic range of image is increased, contrast is expanded, and makes image streak feature more obvious;

Step 5: to calculate in this image position the most clearly, position during the collected image of object corresponding to this position is optimum photographing position.

Step 5.1: obtaining image has fuzzy-clear-fuzzy progressive formation, and stitching image is horizontally divided into multiple little pictures.

Step 5.2: each little picture that step 5.1 obtains has obvious marginal band.To focus good position, namely have the image at more sharp-pointed edge, the gradient function value that Ying Yougeng is large.First use energy gradient function, by the Grad of a Difference Calculation point of consecutive points, this Grad is characterized by the definition numerical value of this position.Obtain the definition values of each little image in this approach, then which little picture the most articulation point of Primary Location is positioned at.As Fig. 3, image does not have enough sharp-pointed crest, so just reach the effect of preliminary focusing, the partial trace that ordinate is greater than 0.9, general 10 little pictures namely, these 10 positions of little picture in stitching image are exactly the articulation point position of Primary Location.

Step 5.3: the image after step 5.2 processes can Primary Location articulation point position, then carries out articulation point to the image of this fraction and accurately locates.Equally first carry out the segmentation of 5.1 steps, then wavelet transformation is used to each little picture, definition can be asked further to image.The degree of clear picture or focusing is primarily of how many decisions of the high fdrequency component in image, and when image is the most clear, the high frequency coefficient energy after wavelet decomposition is maximum.As Fig. 4, because figure has obviously sharp-pointed crest, this crest location, the 6th little picture namely, corresponding image coordinate is exactly articulation point position.

Claims (2)

1. an automatic focusing method for line-scan digital camera, the method comprises:

Step 1: change the distance between shot object and camera, with line-scan digital camera, target image is gathered simultaneously;

Step 2: analyze each two field picture collected, analyzed by the size of the value to image slices vegetarian refreshments, if image is entirely black or entirely white, then image is undesirable, re-starts collection;

Step 3: it is a sub-picture that whole distance change procedure is obtained all image mosaic;

Step 4: gray processing process, denoising, enhancing process are carried out to stitching image;

Step 5: to calculate in this image position the most clearly, position during the collected image of object corresponding to this position is optimum photographing position.

2. the automatic focusing method of a kind of line-scan digital camera as claimed in claim 1, is characterized in that the concrete steps of described step 5 are:

Step 5.1: little picture segmentation is carried out to stitching image;

Step 5.2: adopt energy gradient function to carry out preliminary definition judgment to the little picture that 5.1 split, finds in picture of publishing picture and splits picture more clearly;

Step 5.3: adopt wavelet transformation to find out the pixel that in the segmentation picture that step 5.1 obtains, definition is the highest.