CN109712099B - Sonar image equalization method based on SLIC and adaptive filtering - Google Patents
Sonar image equalization method based on SLIC and adaptive filtering Download PDFInfo
- Publication number
- CN109712099B CN109712099B CN201811474850.XA CN201811474850A CN109712099B CN 109712099 B CN109712099 B CN 109712099B CN 201811474850 A CN201811474850 A CN 201811474850A CN 109712099 B CN109712099 B CN 109712099B
- Authority
- CN
- China
- Prior art keywords
- image
- sonar
- pixel
- sonar image
- slic
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Landscapes
- Measurement Of Velocity Or Position Using Acoustic Or Ultrasonic Waves (AREA)
- Image Processing (AREA)
Abstract
The invention belongs to the technical field of sonar image enhancement, and discloses a sonar image equalization method based on SLIC and adaptive filtering, 1. Filtering speckle noise in a sonar image; 2. dividing the sonar image into irregular image blocks by using a simple linear iterative clustering method; 3. respectively carrying out pixel value normalization on each image block, sequencing the pixel values and obtaining a pixel median; 4. updating the gray value of the image block, and traversing all the image blocks according to the step 3; 5. dividing the sonar image into image blocks with different numbers for multiple times, and respectively repeating the step 3 and the step 4;6. and averaging the equalized images obtained in the step 5 for multiple times to obtain a final gray level equalized image. The technical scheme fully considers the gray level similarity of adjacent super-pixel image blocks and the gray level balance of each super-pixel image block, has good gray level balancing effect, and has obvious effect of improving the sonar image classification accuracy.
Description
Technical Field
The invention relates to the technical field of sonar image enhancement, in particular to a sonar image equalization method based on SLIC and adaptive filtering.
Background
Sonar image classification is an important means for submarine target identification, and can be used in the fields of submarine topography detection, submarine investigation, submarine pipeline tracking, state detection and the like. At present, the effect of sonar image classification is seriously influenced by the unbalanced gray scale problem in a sonar image, so that the problem of the unbalanced gray scale of the sonar image has direct influence on the improvement of the accuracy of the sonar image classification.
Influenced by the complex environment of the sea bottom, acoustic noise and the motion state of a towed body, the sonar image often has gray level distortion. The existing sonar image gray level equalization method is mainly based on the motion states of a towed body, and the states are provided by an attitude sensor and a height sensor in the towed body. At present, a SONAR image correction method is introduced by Capus et al in 2008 paper Data correction for visualization and classification of side scan SONAR image. Influenced by wave motion and complex seabed terrain, the towed body motion state that the sensor gave is in the change state all the time, and simultaneously, the sensitivity and the degree of accuracy of sensor self all have the error, corrects the sonar image with the help of these sensor data, can make the correction process have great calculated amount and calculation error. In the segmentation and classification of sonar images, gray level imbalance is a main problem influencing the accuracy of the segmentation of the sonar images. In the current research on sonar image segmentation and classification, the problem of gray level imbalance is not considered independently, and people mainly focus on a classification or segmentation method, so that although a good classification or segmentation effect can be obtained, the algorithm design is complex, and the process is time-consuming.
Simple Linear Iterative Clustering (SLIC): simple Linear Iterative Clustering.
Disclosure of Invention
The invention aims to obtain a sonar image without gray distortion aiming at the technical problems, provides a sonar image equalization method based on SLIC and adaptive filtering, realizes gray equalization of various sonar images, provides a corrected sonar image for seabed target detection and classification, carries out super-pixel clustering on the sonar images, filters clustered and divided sonar image blocks based on adaptive filtering, and realizes gray equalization of the sonar images by repeating the steps for multiple times. The method comprises the following specific steps:
step 1, filtering speckle noise in a sonar image;
step 2, dividing the sonar image into irregular image blocks by using a simple linear iterative clustering method;
step 3, respectively carrying out pixel value normalization on each image block, sequencing the pixel values and obtaining a pixel median;
for image block P 0 And all its neighboring image blocks { P 1 ,P 2 ,K,P k To uniquely identify their pixel values, resulting in a pixel family { A } 0 ,A 1 ,A 2 ,...,A k }，A i (i =0,1, ·, k) represents an image block P i The result of the pixel value normalization of (i =0,1.., k) removes P i The repeated pixel values of (1); to A i (i =0,1.., k) the pixel values are ordered in order from small to large; get A i Median value of a i (i＝0,1,...,k)。
Step 4, updating the gray value of the image block, and traversing all the image blocks according to the step 3;
the specific method comprises the following steps: if a i (i＝1,2, k) is equal to 0 or 255, then the image block P is i The pixel values in (i =1,2.., k) are all reset to 0 or 255; if a i (i =1,2,. K) is not equal to 0, nor 255, and | a i -a 0 If | i =1,2,.., k) is less than the set threshold, then P is used i '＝a 0 + -rand (X) updating image block P i (i =1,2,.., k) and a rank (X) is indicated at [1,X & (k) ]]Randomly generating a number within the range, and the updated image block is P i '。
Step 5, dividing the sonar image into image blocks with different numbers for multiple times, and respectively repeating the step 3 and the step 4;
and 6, averaging the equalized images obtained in the step 5 for multiple times to obtain a final gray level equalized image.
Further: x =5.
Further: the number of image blocks is 3000.
Further: the number of image blocks in step 5 is 2500 and/or 2000 and/or 1500 and/or 1000.
The technical scheme is that a sonar image is divided into irregular super-pixel image blocks based on a simple linear iterative clustering method, each super-pixel image block is filtered by using a self-adaptive filtering method, the gray level similarity of adjacent super-pixel image blocks and the gray level balance of each super-pixel image block are fully considered, the integral balance of the image is realized through local balance, and after the gray level balance is realized, an ideal segmentation result can be obtained by using a simple image segmentation method (such as a k-means clustering algorithm); the method does not use the attitude data of the towed body carried by sonar, and avoids inherent data errors in the calculation process. The sonar image obtained by the method can solve the problem of unbalanced gray level in the sonar image, has good gray level equalization effect, and has obvious effect of improving the sonar image classification accuracy.
Drawings
FIG. 1: a sonar image;
FIG. 2: simple linear iterative clustering is performed on the clustering result of fig. 1;
FIG. 3: the gray scale equalization results of fig. 1;
FIG. 4: the segmentation result of the graph 1 is subjected to k-means clustering algorithm;
FIG. 5: the k-means clustering algorithm matches the segmentation results of fig. 3.
Wherein the white grid lines in fig. 2 are the boundaries of each cluster block.
Detailed Description
Embodiments of the present invention will be described in detail with reference to fig. 1 to 5.
Step 1, carrying out median filtering on the sonar image, wherein the size of a filter is 3 by 3, and the purpose of the step is to filter out speckle noise in the sonar image. Median filtering is a common method in image filtering.
And 2, dividing the sonar image into irregular image blocks by using a simple linear iterative clustering method (proposed in Achata 2012). The size of the sonar image in FIG. 1 is 419X 317, K l ＝3000，K l Representing the number of image blocks divided by the SLIC. Fig. 2 shows the SLIC clustering results, and the white grid lines in the graph are the boundaries of each clustering block.
And 3-5, using adaptive filtering for each image block obtained by SLIC, wherein the step is a key step of the method. The specific process is as follows:
a. and respectively carrying out pixel value normalization on each image block and adjacent image blocks thereof. For image block P 0 And all its neighboring image blocks { P 1 ,P 2 ,K,P k To uniquely identify their pixel values, resulting in a pixel family { A } 0 ,A 1 ,A 2 ,...,A k }，A i (i =0,1.., k) represents image block P i (i =0,1,.., k) is the result of the normalization of the pixel values. If P i Has a pixel value of {4,7,0,0,0,1,2,3,3,4, the result A of pixel value normalization i Is {4,7,0,1,2,3}.
b. To A i The pixel values in (i =0,1.., k) are ordered from small to large. If A i Is 4,7,0,1,2,3, and the result of the sorting is 0,1,2,3,4,7.
c. After sorting, finding the median a of the pixel values i (i＝0,1,...,k)。A i Median of a i Is 2.5.
The processing of the steps a-c can reduce the influence of gray value imbalance on the median value. If the image block is balanced in gray value distribution, the maximum value and the minimum value of the gray values of the image block are not much different from the median value. If the gray values of the image block are not evenly distributed, the difference between the maximum value and the minimum value may be larger. When the number of smaller gray values is more, the median value is biased to the smaller gray value; when the number of larger gradation values is larger, the median value is biased to a larger gradation value. The uniqueness process in step a can reduce this bias.
d. If a i (i =1,2,. K) equal to 0 or 255, then image block P i The pixel values in (i =1,2.., k) are all reset to 0 or 255.
If a i (i =1,2,. K) is not equal to 0, nor 255, and | a i -a 0 I (i =1,2, ·, k) is smaller than the set threshold, indicating that the image block P is image block P i And image block P 0 Update the image block P according to equation (1) i A gray value, rand (5) of (i =1,2,.. K) is indicated at [1,5]A number is randomly generated within the range. The updated image block is P i '. After update, the image block P i ' Gray value and P 0 Approximately, realize P 0 The neighborhood gray value of (2) is equalized.
P i '＝a 0 ±rand(5) (1)
e. All image blocks are traversed according to steps a-d.
Step 6, use smaller K l Value (e.g. K) l ＝2500，K l ＝2000，K l ＝1500，K l = 1000), repeating steps 3-5 for different K l And averaging the equalized images obtained by the values to obtain a final gray level equalized image. The gray scale equalization results of fig. 1 are shown in fig. 3.
Through gray level equalization, the contrast of a target (a white highlight part in fig. 3) in a sonar image is more obvious, and the problem of gray level imbalance in fig. 1 is effectively solved; after gray level equalization, effective segmentation of the sonar image can be realized by using a simple segmentation method. Fig. 4 and 5 use a simple k-means clustering algorithm to segment the sonar image (fig. 1 and 3) into a target region (white), a background region (gray), and a shadow region (black). From the results, it is clear that the segmentation effect is better through the gray level equalization.
The examples are merely illustrative of the technical solution of the invention and do not limit it in any way; although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that various changes may be made and equivalents may be substituted for elements thereof; such modifications and substitutions do not depart from the spirit and scope of the corresponding claims.
Claims (5)
1. A sonar image equalization method based on SLIC and adaptive filtering is characterized by comprising the following steps:
step 1, filtering speckle noise in a sonar image;
step 2, dividing the sonar image into irregular image blocks by using a simple linear iterative clustering method;
step 3, respectively carrying out pixel value normalization on each image block, sequencing the pixel values and obtaining a pixel median;
step 4, updating the gray value of the image block, and traversing all the image blocks according to the step 3;
step 5, dividing the sonar image into image blocks with different numbers for multiple times, and respectively repeating the step 3 and the step 4;
step 6, averaging the equalized images obtained in the step 5 for multiple times to obtain a final gray level equalized image;
the specific method of the step 3 comprises the following steps: for image block P 0 And all its neighboring image blocks { P 1 ,P 2 ,… ,P k To uniquely identify their pixel values, resulting in a pixel family { A } 0 ,A 1 ,A 2 ,...,A k }，A i (i =0,1.., k) represents image block P i (i =0,1.., k) removing P as a result of the normalization of the pixel values i The repeated pixel values of (1); to A i The pixel values in (i =0,1.., k) are arranged in order from small to largeSequencing; get A i Median of a i (i＝0,1,...,k)；
The specific method of the step 4 comprises the following steps: if a i (i =1,2,. K) equal to 0 or 255, then image block P i The pixel values in (i =1,2.., k) are all reset to 0 or 255; if a i (i =1,2,. K) is not equal to 0, nor 255, and | a i -a 0 If | i =1,2,.., k) is less than the set threshold, then P is used i '＝a 0 + -rand (X) updating image block P i A gray value, rand (X), of (i =1,2,.., k) is indicated at [1,X]Randomly generating a number within the range, and the updated image block is P i '。
2. The sonar image equalization method based on SLIC and adaptive filtering according to claim 1, characterized in that the specific method of step 5 is: the sonar image is divided into a smaller number of image blocks.
3. The sonar image equalization method based on SLIC and adaptive filtering according to claim 1, characterized in that: x =5.
4. The sonar image equalization method based on SLIC and adaptive filtering according to any one of claims 1 or 3, characterized in that: the number of the image blocks in the step 2 is 3000.
5. The sonar image equalization method based on SLIC and adaptive filtering according to claim 4, characterized in that: the number of image blocks in step 5 is 2500 and/or 2000 and/or 1500 and/or 1000.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811474850.XA CN109712099B (en) | 2018-12-04 | 2018-12-04 | Sonar image equalization method based on SLIC and adaptive filtering |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811474850.XA CN109712099B (en) | 2018-12-04 | 2018-12-04 | Sonar image equalization method based on SLIC and adaptive filtering |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109712099A CN109712099A (en) | 2019-05-03 |
| CN109712099B true CN109712099B (en) | 2022-12-27 |
Family
ID=66254028
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811474850.XA Active CN109712099B (en) | 2018-12-04 | 2018-12-04 | Sonar image equalization method based on SLIC and adaptive filtering |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109712099B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113962876A (en) * | 2019-07-17 | 2022-01-21 | 中国电子科技集团公司第十三研究所 | Correction method and correction device for pixel distortion and terminal |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1347621A (en) * | 1999-12-14 | 2002-05-01 | 皇家菲利浦电子有限公司 | Reducing 'blocking picture' effects |
| CN110706177A (en) * | 2019-09-30 | 2020-01-17 | 北京大学 | Method and system for equalizing gray level of side-scan sonar image |
| CN110852959A (en) * | 2019-10-14 | 2020-02-28 | 江苏帝一集团有限公司 | Sonar image filtering method based on novel median filtering algorithm |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102968769B (en) * | 2012-11-27 | 2015-07-22 | 宁波盈芯信息科技有限公司 | Image consistency enhancing device |
| US9589323B1 (en) * | 2015-08-14 | 2017-03-07 | Sharp Laboratories Of America, Inc. | Super resolution image enhancement technique |
| US10419758B2 (en) * | 2017-03-16 | 2019-09-17 | Mediatek Inc. | Non-local adaptive loop filter processing |
-
2018
- 2018-12-04 CN CN201811474850.XA patent/CN109712099B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1347621A (en) * | 1999-12-14 | 2002-05-01 | 皇家菲利浦电子有限公司 | Reducing 'blocking picture' effects |
| CN110706177A (en) * | 2019-09-30 | 2020-01-17 | 北京大学 | Method and system for equalizing gray level of side-scan sonar image |
| CN110852959A (en) * | 2019-10-14 | 2020-02-28 | 江苏帝一集团有限公司 | Sonar image filtering method based on novel median filtering algorithm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109712099A (en) | 2019-05-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113313641B (en) | CT image denoising method with self-adaptive median filtering | |
| Harikiran et al. | Impulse noise removal in digital images | |
| CN109685730B (en) | Wavelet denoising method based on self-adaptive non-local mean value | |
| CN106296763B (en) | A kind of metal material Industry CT Image Quality method for quickly correcting | |
| CN111783583B (en) | SAR image speckle suppression method based on non-local mean algorithm | |
| CN106991661B (en) | Non-local mean denoising method fusing KL (karhunen-Loeve) transformation and grey correlation degree | |
| CN112907460B (en) | Remote sensing image enhancement method | |
| CN107169962B (en) | Gray level image fast segmentation method based on space density constraint kernel fuzzy clustering | |
| CN110222747B (en) | Optimized clustering method | |
| CN109712099B (en) | Sonar image equalization method based on SLIC and adaptive filtering | |
| CN113989168A (en) | Self-adaptive non-local mean filtering method for salt and pepper noise | |
| CN103778611A (en) | Switch weighting vector median filter method utilizing edge detection | |
| CN111429372A (en) | Method for enhancing edge detection effect of low-contrast image | |
| CN109903254A (en) | Based on the improved bilateral filtering method of Poisson's kernel | |
| CN111833275A (en) | Image denoising method based on low-rank analysis | |
| CN109636822B (en) | Improved Canny self-adaptive edge extraction method based on newly-constructed membership function | |
| Kouhi et al. | A modified FCM algorithm for MRI brain image segmentation | |
| CN111340741A (en) | Particle swarm optimization gray level image enhancement method based on quaternion and L1 norm | |
| Kusnik et al. | On the robust technique of mixed Gaussian and impulsive noise reduction in color digital images | |
| CN110288544A (en) | Image de-noising method based on net―function | |
| CN109685729A (en) | A kind of Poisson denoising method of Pixel-level threshold adaptive | |
| CN110533605B (en) | Accurate noise point calibration method | |
| CN113361633A (en) | Medical image big data 3D residual error network classification method | |
| Li et al. | Novel approach to pavement cracking detection based on morphology and mutual information | |
| CN111861947A (en) | Method and device for improving information entropy of histogram technology enhanced image |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |