CN101295401A - Infrared point target detecting method based on linear PCA - Google Patents
Infrared point target detecting method based on linear PCA Download PDFInfo
- Publication number
- CN101295401A CN101295401A CNA200810038548XA CN200810038548A CN101295401A CN 101295401 A CN101295401 A CN 101295401A CN A200810038548X A CNA200810038548X A CN A200810038548XA CN 200810038548 A CN200810038548 A CN 200810038548A CN 101295401 A CN101295401 A CN 101295401A
- Authority
- CN
- China
- Prior art keywords
- msub
- image
- mrow
- target
- sub
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims description 23
- 238000001514 detection method Methods 0.000 claims abstract description 61
- 238000012549 training Methods 0.000 claims abstract description 25
- 238000000513 principal component analysis Methods 0.000 claims description 28
- 239000013598 vector Substances 0.000 claims description 22
- 239000004576 sand Substances 0.000 claims description 2
- 238000001914 filtration Methods 0.000 abstract description 12
- 238000007781 pre-processing Methods 0.000 abstract 1
- 238000005516 engineering process Methods 0.000 description 4
- 238000013144 data compression Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000005764 inhibitory process Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000000877 morphologic effect Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000012567 pattern recognition method Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
Images
Landscapes
- Image Processing (AREA)
- Photometry And Measurement Of Optical Pulse Characteristics (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a detection method based on a linear PCA for infrared point targets, comprising the following steps of: first, creating a training sample of the infrared point target through an improved gaussian gray model, and carrying out trainings for the PCA through the samples, to produce a group of principal components; then intercepting and acquiring sub-images from an infrared image to be detected, and respectively projecting the sub-images to each principal component, to reconstruct the sub-images through obtained projection coefficients and the principal components, and calculate the reconstruction error; finally, substituting a measure function with the reconstruction error, to produce a detection image and improve the testability of the target. The detection method based on the linear PCA for infrared point targets can greatly improve the testability of the infrared point targets. Different from the filtration-based detection method, the detection method based on the linear PCA for infrared point targets does not require a preprocessing for the infrared images to be detected.
Description
Technical Field
The invention relates to an infrared target detection method in the technical field of image processing, in particular to an infrared point target detection method based on linear PCA.
Background
In a video monitoring and searching system, the application of the infrared camera technology has great particularity, can work all weather, is a passive and passive detection technology, and can well conceal oneself when finding a target. Among them, how to detect the target under the remote condition is one of the key technologies of the infrared monitoring and searching system. If the target can be detected at a longer distance, the system has more time to react to the target, and the system is provided for active position in confrontation. In addition, target detection is the first step in the system's reaction to a target, and the performance of target detection directly affects the effect of subsequent processing (such as target tracking and target recognition). Therefore, the infrared target detection technology is significant for an infrared video system.
When the target is far from the infrared camera system, the area of the target on the imager is small, and the target is a small point-like target (generally less than 100 pixels). Due to the influences of factors such as low contrast, small area, fuzzy texture and edge, strong noise and clutter interference and the like of the infrared point target, the detection of the infrared point target is very difficult. Most of traditional infrared point target detection methods are based on filters, and the methods based on spatial filters include: high-pass template filtering, median filtering, mathematical morphology filtering, local standard deviation filtering, and the like; the method based on the frequency domain filter comprises the following steps: ideal high-pass filtering, Butterworth high-pass filtering, Gaussian high-pass filtering and the like. If the image data (a vector in which the gradation values of all pixels in a sub-image centered on each pixel in the image are arranged) is regarded as a data set composed of the object class data and the background class data, the problem of detecting an object from the image is converted into a problem of identifying (classifying) the object data from the image data. Thus, some pattern recognition methods may be used to achieve target detection. PCA (principal component analysis) is an important method in pattern recognition theory, and is often used to implement data compression. The feature vectors (principal components ) that can describe the training data most, that is, the feature vectors corresponding to the largest several feature values, are obtained through training. The PCA also extracts the features of the training data while realizing data compression.
Through the search of the prior art documents, the Chinese application number: 200410068024.7 patent name "morphological filter automatic target detection method", this patent is based on the morphological filter method detection target of Top-hat operator, the concrete principle is: the small target is in a high frequency band, and the background is in a low frequency band, so that the purpose of highlighting the target and suppressing the background can be realized by high-frequency filtering or subtracting an image obtained by low-pass filtering from an original image, and target detection is finished. Obviously, in this method, some noise points (also in the high frequency band) have larger output while the target is detected. Thus, noise points can be falsely detected as targets, and the false alarm rate is increased, which is an inherent defect of the infrared point target detection method based on the filter.
Disclosure of Invention
The invention provides an infrared point target detection method based on linear PCA (principal component analysis), aiming at solving the problems in the prior art, and the detection of infrared point targets of various wave bands can be realized.
The invention is realized by the following technical scheme, and the method comprises the following steps:
firstly, generating training samples of an infrared point target by using an improved Gaussian gray model, and training PCA (principal component analysis) through the samples to generate a group of principal components;
then, intercepting a sub-image from the infrared image to be detected, respectively projecting the sub-image onto each principal component, reconstructing the sub-image by using the obtained projection coefficient and the principal component, and calculating a reconstruction error;
and finally, substituting the reconstruction error into a detection function to generate a detection image, so that the detectability of the target is improved.
The improved Gaussian gray model generates an infrared point target training sample, and specifically comprises the following steps: in the conventional Gaussian gray model
<math>
<mrow>
<mi>I</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msub>
<mi>I</mi>
<mi>max</mi>
</msub>
<mi>exp</mi>
<mrow>
<mo>(</mo>
<mo>-</mo>
<mfrac>
<mn>1</mn>
<mn>2</mn>
</mfrac>
<mrow>
<mo>[</mo>
<mfrac>
<msup>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>-</mo>
<msub>
<mi>x</mi>
<mn>0</mn>
</msub>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<msubsup>
<mi>σ</mi>
<mi>x</mi>
<mn>2</mn>
</msubsup>
</mfrac>
<mo>+</mo>
<mfrac>
<msup>
<mrow>
<mo>(</mo>
<mi>j</mi>
<mo>-</mo>
<msub>
<mi>y</mi>
<mn>0</mn>
</msub>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<msubsup>
<mi>σ</mi>
<mi>y</mi>
<mn>2</mn>
</msubsup>
</mfrac>
<mo>]</mo>
</mrow>
<mo>)</mo>
</mrow>
</mrow>
</math>
On the basis of (2), adding two constraint conditions
<math>
<mrow>
<msub>
<mi>T</mi>
<mn>1</mn>
</msub>
<mo>≤</mo>
<mfrac>
<msub>
<mi>I</mi>
<mi>max</mi>
</msub>
<msub>
<mi>σ</mi>
<mi>x</mi>
</msub>
</mfrac>
<mo>≤</mo>
<msub>
<mi>T</mi>
<mn>2</mn>
</msub>
</mrow>
</math>
And
<math>
<mrow>
<msub>
<mi>T</mi>
<mn>1</mn>
</msub>
<mo>≤</mo>
<mfrac>
<msub>
<mi>I</mi>
<mi>max</mi>
</msub>
<msub>
<mi>σ</mi>
<mi>y</mi>
</msub>
</mfrac>
<mo>≤</mo>
<msub>
<mi>T</mi>
<mn>2</mn>
</msub>
<mo>,</mo>
</mrow>
</math>
make it an improved gaussian gray model:
thus, the improvement of the gaussian gray scale model is completed. Wherein, ImaxIs the target center pixel value (gray peak); sigmaxFor the horizontal spread parameter, σyControlling the walking characteristics of the target pixel for the vertical scattering parameter; (x)0，y0) The central coordinates of the target image are taken; (i, j) are other pixel coordinates of the target image. I ismax，σxAnd σyThree parameters determine the characteristics of the gaussian gray scale model. The traditional infrared point target generated by the Gaussian gray scale model can generate some 'point targets' which cannot be generated in the actual infrared image, for example, some point targets similar to noise points (when I ismaxIs large, and σxAnd σySmaller), if such training samples (samples) are usedThis Outlier) to train the PCA, the detection performance is greatly reduced.
The PCA is trained with the sample to generate a set of principal components, specifically: and generating a plurality of infrared point target images by using the improved Gaussian gray model, wherein the images are used as training samples. And (3) arranging the pixel gray values of the generated image samples of each infrared point target into vectors according to the head phase of a line, training PCA by using the vectors to generate principal components, and taking n characteristic vectors with the maximum corresponding characteristic values to finish target detection. The value of n is determined by taking the detection effect and the computational complexity (magnitude of dimensionality reduction) into consideration, and is preferably equal to 6.
Intercepting a sub-image from an infrared image to be detected, respectively projecting the sub-image onto each principal component, reconstructing the sub-image by using the obtained projection coefficient and the principal component, and calculating a reconstruction error, wherein the method specifically comprises the following steps:
step one, intercepting a sub-image P at an (x, y) position in a detected imagesAnd generates a vector PscA1 is to PscRespectively projected on n main components to obtain n projection coordinates with a formula of
Wherein, ω iskIs PscAt the k-th feature vector
The projection coordinate of (a) is an average target;
reconstructing the sub-image by using n principal components and n projection coordinates to obtain a reconstructed image, wherein the formula is
Wherein,
to reconstruct an image;
step three, calculating the reconstruction error of the subimage with the formula
Wherein epsilonRCTo reconstruct the error. If the sub-picture PsIs the target image, then the image is reconstructed
And PsThe similarity is very large, otherwise the similarity is very small, so that the reconstruction error is a measure of whether the sub-image is the target image, and the smaller value of the reconstruction error corresponds to the existence of the target.
Substituting the reconstruction error into the detection function to generate a detection image, which specifically comprises the following steps: will reconstruct the error εRCSubstituting a detection function
<math>
<mrow>
<msub>
<mi>ζ</mi>
<mi>D</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msup>
<mi>e</mi>
<mrow>
<mo>-</mo>
<msub>
<mi>ϵ</mi>
<mi>RC</mi>
</msub>
<msup>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mo>/</mo>
<mn>2</mn>
<msup>
<mi>δ</mi>
<mn>2</mn>
</msup>
</mrow>
</msup>
<mo>.</mo>
</mrow>
</math>
Unlike the reconstruction error, ζDA large value of (x, y) corresponds to the presence of a target, which is consistent with human intuition; and ζDThe value of (x, y) is between 0 and 1, so that a normalization link is omitted, and detection indexes such as signal-to-noise ratio (SNR) and the like can be calculated more conveniently. The value of the parameter δ is empirically chosen to be δ 0.25. The detection function value zeta corresponding to each pixel positionD(x, y), as the gradation value of the pixel, the detection image is generated.
Most of traditional infrared point target detection is based on a filtering method, and when a target is detected, a noise point is misjudged as a target point, so that the false alarm rate is improved, and the load of a system is increased. The invention uses a mode identification method to detect an infrared point target, PCA is an important branch of the mode identification and is often used for realizing data compression, but the PCA also extracts the characteristics of the target while compressing the data, and the PCA is used for realizing the detection of the infrared point target by utilizing the characteristic and substantially uses the PCA to 'identify' the point target in a detected image. For the identification method, the influence of the training samples on the identification rate is great, and the method firstly improves the Gaussian gray model and then uses the improved Gaussian gray model to generate the training samples of the infrared point target. And (3) generating principal components through PCA training, respectively projecting detected sub-images (intercepted from detected images) onto each principal component to obtain corresponding projection coordinates, and calculating a reconstruction error through the reconstruction of the projection coordinates and the principal components on the sub-images, wherein the smaller reconstruction error corresponds to the existence of the target. A detection function is designed to produce a detection image in which the detectability of an infrared point target is greatly improved.
The method can greatly improve the detectability of the infrared point target, and is different from a filtering-based method, and the method does not need to preprocess the infrared to-be-detected image.
Drawings
FIG. 1 is a training sample of an infrared point target generated by an improved Gaussian gray scale model
FIG. 2 is a schematic diagram of an embodiment of the present invention
Wherein: (a)1)、(b1) Two original infrared images containing 10 point targets respectively (b)1) Target 6 in (1) is a trueOther targets in the two images are obtained by embedding simulation targets generated by the improved Gaussian gray model into the infrared image, all targets in the image are weak targets, and the detection difficulty is high; (a)2)、(b2) Are each a PCA pair (a)1)、(b1) The result of the detection; (a)3)、(b3) Are respectively (a)2)、(b2) 3D structure diagram of (1).
Detailed Description
The technical scheme of the invention is explained in detail by combining the specific embodiments as follows: the present embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation manner and a specific operation process are given, but the protection scope of the present invention is not limited to the following embodiments.
The embodiment mainly comprises two aspects, namely a method for generating an infrared point target training sample and a PCA-based infrared point target detection method.
As shown in fig. 1, which is a training sample of an infrared point target generated by an improved gaussian gray scale model, wherein: the size of the training sample is 11 × 11, and the training sample is obtained by rotating an image generated by the improved gaussian gray model by an arbitrary angle.
Training
1) Arranging the pixel gray values of 121 target training sample sub-images (11 multiplied by 11) according to the sequence of head-to-tail connection to generate a column vector (121 dimension);
2) arranging the column vectors to generate a matrix, decomposing the characteristic values of the matrix to obtain the optimal characteristic vector for describing the target, and arranging the characteristic vectors according to the characteristic values from large to small; this embodiment takes the first 6 feature vectors
These 6 feature vectors are the set of feature vectors that best describe the target.
(II) detection
3) Intercepting a sub-image to be detected at each pixel position in the detected image by using a sliding window with the size being (11 multiplied by 11) the same as that of the target sample image, and converting the sub-image to be detected into a column vector (121 dimensions);
4) projecting the column vector generated by the sub-image to be detected onto 6 eigenvectors to obtain 6 projection coordinates { omega [ omega ] correspondingly1，ω2，…ωn}；
5) Reconstructing the sub-image to be detected by using the 6 eigenvectors and the projection coordinates, and calculating the reconstruction error epsilonRC；
6) Will epsilonRCSubstituting the detection function ζD(x, y), calculating the detection function value at each pixel position, and generating the detection image by taking the function values as the gray value of the corresponding pixel.
FIG. 2 shows the results of the detection in example (a)2)、(b2) Are each a PCA pair (a)1)、(b1) The result of the detection; (a)3)、(b3) Are respectively (a)2)、(b2) 3D structure diagram of (1). Table 1 gives the detection indices of fig. 2, signal to noise ratio gain (SNRG) and Background Suppression Factor (BSF), respectively. The signal-to-noise ratio gain is a measure of the signal-to-noise ratio improvement capability, and the calculation formula is SNRG ═ SNRout/SNRinSNR inoutTo detect the signal-to-noise ratio, SNR, of the images ((a 2) and (b2) in FIG. 2)inSignal-to-noise ratio of the original image ((a 1) and (b1) in fig. 2); the background inhibition factor is a measure of the background inhibition ability, and the calculation formula is BSF ═ Cin/CoutWherein C isinStandard deviation of the background of the original image, CoutTo detect the standard deviation of the image background. The larger the two detection indexes are, the better the detection performance of the detection method is. As can be seen from Table 1, the signal-to-noise ratio gain obtained by the method of the present invention is greater than 1 except for a few targets in FIG. 2(b1), indicating that the method can improve the detectability of infrared targets. In addition, for the two images of the example, the background was suppressedThe production factors are all far greater than 1, which indicates that the method has strong background inhibition capability.
TABLE 1 comparison of the performance index of target detection in FIG. 2 by three subspace algorithms
Claims (5)
1. An infrared point target detection method based on linear PCA is characterized by comprising the following steps:
firstly, generating training samples of an infrared point target by using an improved Gaussian gray model, and training PCA (principal component analysis) through the samples to generate a group of principal components;
then, intercepting a sub-image from the infrared image to be detected, respectively projecting the sub-image onto each principal component, reconstructing the sub-image by using the obtained projection coefficient and the principal component, and calculating a reconstruction error;
finally, substituting the reconstruction error into a detection function to generate a detection image;
the improved Gaussian gray model specifically comprises the following steps:
wherein, ImaxIs the target center pixel value; sigmaxFor the horizontal spread parameter, σyControlling the walking characteristics of the target pixel for the vertical scattering parameter; (x)0，y0) The central coordinates of the target image are taken; (I, j) is the other pixel coordinate of the target image, Imax，σxAnd σyThree parameters determine the characteristics of the gaussian gray scale model.
2. The method of claim 1, wherein the PCA is trained using a sample to generate a set of principal components, specifically: generating a plurality of infrared point target images by using an improved Gaussian gray model, taking the images as training samples, arranging the pixel gray values of the generated image samples of each infrared point target into vectors according to the head phase of a line, training PCA (principal component analysis) by using the vectors to generate principal components, and taking n characteristic vectors with the maximum corresponding characteristic values to finish target detection.
3. The linear PCA-based infrared point target detection method of claim 2 wherein the value of n is 6.
4. The infrared point target detection method based on linear PCA of claim 1, wherein said method intercepts the sub-images from the infrared image to be detected and projects them onto the principal components, reconstructs the sub-images using the obtained projection coefficients and principal components, and calculates the reconstruction error, comprising the steps of:
step one, intercepting a sub-image P at an (x, y) position in a detected imagesAnd generates a vector PscA1 is to PscRespectively projected on n main components to obtain n projection coordinates with a formula of
Wherein, ω iskIs PscAt the k-th feature vectorThe projection coordinate of (a) is an average target;
reconstructing the sub-image by using n principal components and n projection coordinates to obtain a reconstructed image, wherein the formula is
Wherein,to reconstruct an image;
step three, calculating the reconstruction error of the subimage with the formulaWherein epsilonRCFor reconstruction errors, if the sub-picture PsIs the target image, then the image is reconstructedAnd PsIs similar toIf the image size is large, the similarity is small, and the reconstruction error is a measure of whether the sub-image is the target image.
5. The method of claim 1, wherein the step of generating the detection image by substituting the reconstruction error into the detection function comprises: will reconstruct the error εRCSubstituting a detection function
<math>
<mrow>
<msub>
<mi>ζ</mi>
<mi>D</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msup>
<mi>e</mi>
<mrow>
<mo>-</mo>
<msub>
<mi>ϵ</mi>
<mi>RC</mi>
</msub>
<msup>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mo>/</mo>
<mn>2</mn>
<msup>
<mi>δ</mi>
<mn>2</mn>
</msup>
</mrow>
</msup>
<mo>,</mo>
</mrow>
</math>
The value of the parameter δ is set to 0.25, and the detection function value ζ corresponding to each pixel position is setD(x, y), as the gradation value of the pixel, the detection image is generated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA200810038548XA CN101295401A (en) | 2008-06-05 | 2008-06-05 | Infrared point target detecting method based on linear PCA |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA200810038548XA CN101295401A (en) | 2008-06-05 | 2008-06-05 | Infrared point target detecting method based on linear PCA |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101295401A true CN101295401A (en) | 2008-10-29 |
Family
ID=40065663
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA200810038548XA Pending CN101295401A (en) | 2008-06-05 | 2008-06-05 | Infrared point target detecting method based on linear PCA |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101295401A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101799875A (en) * | 2010-02-10 | 2010-08-11 | 华中科技大学 | Target detection method |
| CN101882314A (en) * | 2010-07-20 | 2010-11-10 | 上海交通大学 | Infrared small target detection method based on overcomplete sparse representation |
| CN101957993A (en) * | 2010-10-11 | 2011-01-26 | 上海交通大学 | Adaptive infrared small object detection method |
| CN105023255A (en) * | 2015-08-05 | 2015-11-04 | 西安电子科技大学 | Infrared weak small target image sequence simulation method for jittering of Gaussian model area-array camera |
| CN106643731A (en) * | 2016-12-29 | 2017-05-10 | 凌云光技术集团有限责任公司 | System and method for tracking and measuring point target |
| CN108320024A (en) * | 2017-01-16 | 2018-07-24 | 佳能株式会社 | Dictionary creating apparatus and method, apparatus for evaluating and method and storage medium |
| CN109511630A (en) * | 2017-09-20 | 2019-03-26 | 南京理工大学 | Intelligent sprayer with target following function |
| CN109741396A (en) * | 2018-12-12 | 2019-05-10 | 天津津航技术物理研究所 | A kind of extremely small and weak infrared target detection method |
-
2008
- 2008-06-05 CN CNA200810038548XA patent/CN101295401A/en active Pending
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101799875B (en) * | 2010-02-10 | 2011-11-30 | 华中科技大学 | Target detection method |
| CN101799875A (en) * | 2010-02-10 | 2010-08-11 | 华中科技大学 | Target detection method |
| CN101882314A (en) * | 2010-07-20 | 2010-11-10 | 上海交通大学 | Infrared small target detection method based on overcomplete sparse representation |
| CN101882314B (en) * | 2010-07-20 | 2012-06-20 | 上海交通大学 | Infrared small target detection method based on overcomplete sparse representation |
| CN101957993A (en) * | 2010-10-11 | 2011-01-26 | 上海交通大学 | Adaptive infrared small object detection method |
| CN105023255B (en) * | 2015-08-05 | 2018-01-30 | 西安电子科技大学 | The Infrared DIM-small Target Image sequence emulation mode of Gauss model area array cameras shake |
| CN105023255A (en) * | 2015-08-05 | 2015-11-04 | 西安电子科技大学 | Infrared weak small target image sequence simulation method for jittering of Gaussian model area-array camera |
| CN106643731A (en) * | 2016-12-29 | 2017-05-10 | 凌云光技术集团有限责任公司 | System and method for tracking and measuring point target |
| CN106643731B (en) * | 2016-12-29 | 2019-09-27 | 凌云光技术集团有限责任公司 | A kind of pair of point target carries out the system and method for tracking measurement |
| CN108320024A (en) * | 2017-01-16 | 2018-07-24 | 佳能株式会社 | Dictionary creating apparatus and method, apparatus for evaluating and method and storage medium |
| CN108320024B (en) * | 2017-01-16 | 2022-05-31 | 佳能株式会社 | Dictionary creation device and method, evaluation device and method, and storage medium |
| US11521099B2 (en) | 2017-01-16 | 2022-12-06 | Canon Kabushiki Kaisha | Dictionary generation apparatus, evaluation apparatus, dictionary generation method, evaluation method, and storage medium for selecting data and generating a dictionary using the data |
| CN109511630A (en) * | 2017-09-20 | 2019-03-26 | 南京理工大学 | Intelligent sprayer with target following function |
| CN109511630B (en) * | 2017-09-20 | 2021-09-21 | 南京理工大学 | Intelligent sprayer with target tracking function |
| CN109741396A (en) * | 2018-12-12 | 2019-05-10 | 天津津航技术物理研究所 | A kind of extremely small and weak infrared target detection method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109636742B (en) | Mode conversion method of SAR image and visible light image based on countermeasure generation network | |
| CN101295401A (en) | Infrared point target detecting method based on linear PCA | |
| CN108389188B (en) | Sparse hyperspectral abnormal target detection method | |
| Zheng et al. | An advanced image fusion algorithm based on wavelet transform: incorporation with PCA and morphological processing | |
| CN104102913B (en) | Wrist vena identification system | |
| CN109919960B (en) | Image continuous edge detection method based on multi-scale Gabor filter | |
| CN104408700A (en) | Morphology and PCA (principal component analysis) based contourlet fusion method for infrared and visible light images | |
| US20060245500A1 (en) | Tunable wavelet target extraction preprocessor system | |
| CN104794440B (en) | A kind of false fingerprint detection method based on the multiple dimensioned LBP of more piecemeals | |
| Ertürk et al. | Hyperspectral image classification using empirical mode decomposition with spectral gradient enhancement | |
| CN109658351A (en) | The high spectrum image denoising method that a kind of combination L0 gradient constraint and local low-rank matrix are restored | |
| CN102005037A (en) | Multimodality image fusion method combining multi-scale bilateral filtering and direction filtering | |
| CN106056523B (en) | Blind checking method is distorted in digital picture splicing | |
| CN105574515A (en) | Pedestrian re-identification method in zero-lap vision field | |
| CN101882314A (en) | Infrared small target detection method based on overcomplete sparse representation | |
| CN102332084A (en) | Identity identification method based on palm print and human face feature extraction | |
| Zhang et al. | Comparison of wavelet, Gabor and curvelet transform for face recognition | |
| CN101957993A (en) | Adaptive infrared small object detection method | |
| CN106897999A (en) | Apple image fusion method based on Scale invariant features transform | |
| CN103295221A (en) | Water surface target motion detecting method simulating compound eye visual mechanism and polarization imaging | |
| CN103870836A (en) | POCS (Projections Onto Convex Sets) super-resolution reconstruction-based SAR (Synthetic Aperture Radar) image target recognition method | |
| Jing et al. | SAR image de-noising via grouping-based PCA and guided filter | |
| Zhang et al. | Salient region detection in remote sensing images based on color information content | |
| CN103324921B (en) | A kind of mobile identification method based on interior finger band and mobile identification equipment thereof | |
| CN101430760B (en) | Human face super-resolution processing method based on linear and Bayesian probability mixed model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Open date: 20081029 |