An archive image sharpening method based on local invariant features of images

Abstract

The application discloses a kind of based on image local invariant feature's archive image clear method, including first obtaining the archive image needing to carry out clear, preliminary pre-processing operation is carried out to the archive image to be handled, the archive image region after pre-processing is divided, the image local invariant feature is extracted to each division small area, these feature parameters are saved with area serial number and different feature parameters as benchmark, corresponding determination is carried out to image contrast, light and dark degree, background figure spot and background color, whether archive image can reach the requirement of image contrast, light and dark degree, archive character, background figure spot and background color is observed.The application can automatically find optimal processing parameter, is favorable to realize archive image clear automation processing, simultaneously is favorable to the integration of solution, can be integrated with current archive processing system, realizes archive image clear processing module, completes clear processing task.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of image technology, in particular to a file image sharpening method based on local invariant features of images. BACKGROUND

[0002] Paper file imaging is a digital work for old file storage at present, which has been started for many years. Due to the quality problem of early scanning equipment, part of the scanned file images presents unclear phenomenon. In addition, even now, due to the non-standard shooting or poor shooting conditions during scanning files, part of the current scanned file images also presents unclear phenomenon. These unclear phenomena need to be processed in the later stage to meet the requirements of file image storage.

[0003] There are many solutions to the problem of unclear file images at present, but the effect of some methods is not ideal. In addition, some methods require manual intervention for file image processing, which is troublesome, time-consuming and laborious. Therefore, the file image sharpening method is not ideal, and the existing technology still lacks an automatic processing method for file image sharpening, which cannot achieve the purpose of effectively solving the file image sharpening. SUMMARY

[0004] The purpose of the present application is to provide a file image sharpening method based on local invariant features of images to solve the problems in the background technology.

[0005] To achieve the above purpose, the present application provides the following technical solution: a file image sharpening method based on local invariant features of images, comprising the following steps:

[0006] Step 1: first, acquire the file image that needs to be sharpened;

[0007] Step 2: perform preliminary preprocessing operation on the file image to be processed;

[0008] Step 3: divide the file image region after preprocessing;

[0009] Step 4: extract local invariant features of images for each small region, and save these feature parameters based on region serial number and different feature parameters;

[0010] Step 5: make corresponding determination on image contrast, brightness, file text, background spot and background color;

[0011] Step 6: observe whether the file image meets the requirements of image contrast, brightness, file text, background spot and background color, and if not, perform file image sharpening processing;

[0012] Step seven: first, according to the above-mentioned image sharpening processing method, a moderate parameter is selected for the first image processing;

[0013] Step eight: after the first image processing is completed, the image local invariant features are extracted for each divided small area, and then the correlation analysis is performed on the extracted feature parameters and the feature parameters extracted in step four to obtain the correlation determination degree;

[0014] Step nine: the parameters of each image processing method used in step seven are increased and decreased, the same feature extraction operation is performed, and the correlation analysis is performed on the initial feature parameters to obtain the correlation determination degree;

[0015] Step ten: compare the three correlation determination degrees, select the highest correlation determination degree, and select the highest correlation parameter;

[0016] Step eleven: the parameters of each image processing method in step ten are increased and decreased, the same feature extraction operation is performed, and the correlation analysis is performed on the feature parameters extracted in step four to obtain the correlation determination degree;

[0017] Step twelve: compare the three correlation determination degrees, select the highest correlation determination degree, and select the highest correlation parameter;

[0018] Step thirteen: repeat the operations of step eleven and step twelve until the parameter increase and decrease ratio is reduced to within 5%, and finally select the highest correlation parameter as the sharpening processing parameter;

[0019] Step fourteen: use the sharpening processing parameter in step thirteen to perform formal sharpening processing on the file image to obtain the sharpened image.

[0020] Preferably, the region division in step three is 100*100 pixels, and the image local invariant features in step three are corner points, shift feature points and MESR feature points.

[0021] Preferably, the corresponding determination requirements in step five are moderate contrast, moderate light and dark degree, file character recognition character quantity ratio cannot be lower than 85%, background spot quantity cannot exceed threshold value, and background color presents white, and the background spot quantity according to dark background spot area ratio cannot exceed 5%.

[0022] Preferably, the image sharpening processing method comprises adjusting image contrast, light and dark degree, background extraction processing and image opening and closing operation on file character table.

[0023] Preferably, the respective image processing method in step nine increases and decreases the parameter setting by 50%, and the respective image processing method in step eleven increases and decreases the parameter setting by 25%.

[0024] Technical effects and advantages of the present application:

[0025] The present application can automatically find the optimal processing parameters, is conducive to realizing the automatic processing of the clear archive image, and is conducive to the integration of the solution, can be integrated with the current archive processing system, realizes the archive image clear processing module, and completes the clear processing task. BRIEF DESCRIPTION OF DRAWINGS

[0026] Figure 1 The figure is a system diagram of the archive image clear method of the image local invariant feature of the present application. DETAILED DESCRIPTION

[0027] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0028] The present application provides an archive image clear method based on an image local invariant feature as shown in Figure 1 The archive image clear method based on an image local invariant feature comprises the following steps.

[0029] Step one: first, an archive image needing to be clear is acquired;

[0030] Step two: a preliminary preprocessing operation is performed on the archive image to be processed;

[0031] Step three: the archive image region after preprocessing is divided;

[0032] Step four: image local invariant features are extracted for each small divided region, and these feature parameters are saved with the region serial number and different feature parameters as the basis;

[0033] Step five: corresponding determination is performed on the image contrast, brightness, archive text, background image spot and background color, etc.

[0034] Step six: whether the archive image can meet the requirements of the image contrast, brightness, archive text, background image spot and background color is observed, and if not, archive image clear processing is performed;

[0035] Step seven: a moderate parameter is selected according to the above image clear processing method to perform first image processing;

[0036] Step eight: after the first image processing is completed, the local invariant features of the image are extracted for each divided small area, and then the correlation analysis is performed on the extracted feature parameters and the feature parameters extracted in step four to obtain a correlation determination degree;

[0037] Step nine: the parameters of each image processing method used in step seven are increased and decreased, the same feature extraction operation is performed, and the correlation analysis is performed on the initial feature parameters to obtain a correlation determination degree;

[0038] Step ten: compare the three correlation determination degrees, select the highest correlation determination degree, and select the highest correlation parameter;

[0039] Step eleven: the parameters of each image processing method in step ten are increased and decreased, the same feature extraction operation is performed, and the correlation analysis is performed on the feature parameters extracted in step four to obtain a correlation determination degree;

[0040] Step twelve: compare the three correlation determination degrees, select the highest correlation determination degree, and select the highest correlation parameter;

[0041] Step thirteen: repeat the operations of step eleven and step twelve until the parameter increase and decrease ratio is reduced to within 5%, and finally select the highest correlation parameter as the sharpening processing parameter;

[0042] Step fourteen: use the sharpening processing parameter in step thirteen to perform formal sharpening processing on the file image to obtain a sharpened image;

[0043] The parameters of each image processing method in step eight are increased and decreased, the same feature extraction operation is performed, and the correlation analysis is performed on the initial feature parameters to obtain a correlation determination degree;

[0044] Compare the three correlation determination degrees, select the highest correlation determination degree, and select the highest correlation parameter;

[0045] The parameters of each image processing method in step ten are increased and decreased, the same feature extraction operation is performed, and the correlation analysis is performed on the initial feature parameters to obtain a correlation determination degree;

[0046] Compare the three correlation determination degrees, select the highest correlation determination degree, and select the highest correlation parameter;

[0047] Repeat the operations of step eleven and step twelve until the parameter increase and decrease ratio is reduced to within 5%, and finally select the highest correlation parameter as the sharpening processing parameter;

[0048] The formal clear image is obtained by using the optimal processing parameters, and then the next archive image to be processed can be automatically processed, and the unclear archive image can be processed, so that the archive image is clear.

[0049] The region in step three is 100*100 pixels, and the local invariant features in step three are corner points, shift feature points and MESR feature points.

[0050] The corresponding judgment requirements in step five are that the contrast is moderate, the light and dark degree is moderate, the number of archive text recognition text cannot be less than 85%, the number of background spots cannot exceed the threshold, and the background color is white, and the number of background spots cannot exceed 5% according to the dark background spot area ratio.

[0051] The image clear processing method in step six includes adjusting the image contrast, light and dark degree, background extraction processing and image opening and closing operation on the archive text table.

[0052] The improvement and reduction parameter settings of each image processing method in step nine are 50%, and the improvement and reduction parameter settings of each image processing method in step eleven are 25%.

[0053] Finally, it should be noted that: the above only for the preferred embodiments of the present application, and not for limiting the present application, although the foregoing embodiments of the present application are described in detail, for those skilled in the art, it still can be modified, or part of the technical features of the equivalent replacement, within the spirit and principles of the present application, any modification, equivalent replacement, improvement, etc., should be included in the protection scope of the present application.

Claims

1. A method for enhancing archival images based on local invariant features, characterized in that, Includes the following steps: Step 1: First, obtain the archival image that needs to be sharpened; Step 2: Perform preliminary preprocessing operations on the archive images to be processed; Step 3: Divide the preprocessed archive image regions; Step 4: Extract local invariant features from each subdivided region and save these feature parameters based on region number and different feature parameters; Step 5: Make appropriate judgments on image contrast, brightness, document text, background patterns, and background color; Step Six: Observe whether the archive image meets the requirements for image contrast, brightness, archive text, background patterns and background color. If it does not meet the requirements, perform archive image sharpening processing. The image sharpening processing method includes adjusting image contrast, brightness, background extraction processing and performing image opening and closing operations on archive text tables. Step 7: First, select an appropriate parameter based on the image sharpening method described above and perform the first image processing. Step 8: After the first image processing is completed, extract local invariant features for each divided small region, and then perform correlation analysis on the feature parameters extracted in this step and the feature parameters extracted in step 4 to obtain the degree of correlation determination. Step 9: Increase or decrease the parameters of each image processing method used in Step 7, perform the same feature extraction operation, and conduct correlation analysis with the initial feature parameters to obtain the degree of correlation determination; Step 10: Compare the three correlation levels and select the one with the highest correlation level; Step 11: Increase or decrease the parameters of each image processing method in Step 10, perform the same feature extraction operation, and conduct correlation analysis with the feature parameters extracted in Step 4 to obtain the degree of correlation determination. Step 12: Compare the three correlation levels and select the one with the highest correlation level; this is the parameter with the highest correlation. Step 13: Repeat steps 11 and 12 until the parameter increase and decrease ratio drops to within 5%, and finally select the parameter with the highest correlation as the sharpening parameter; Step Fourteen: Use the sharpening parameters from Step Thirteen to perform formal sharpening processing on the archive image to obtain the sharpened image.

2. The method for enhancing archival images based on local invariant features according to claim 1, characterized in that, The region in step three is divided into 100*100 pixels, and the local invariant features of the image in step three are corner points, Shift feature points, and MESR feature points.

3. The method for enhancing archival images based on local invariant features according to claim 1, characterized in that, The corresponding judgment requirements in step five are: moderate contrast, moderate brightness, the ratio of the number of texts in the document text recognition is not less than 85%, the number of background patches does not exceed the threshold, and the background color is white, and the number of background patches does not exceed 5% based on the area ratio of dark background patches.

4. The method for enhancing archival images based on local invariant features according to claim 1, characterized in that, In step nine, the parameters for each image processing method are increased and decreased by 50%, and in step eleven, the parameters for each image processing method are increased and decreased by 25%.

Images