A face feature recognition system based on digital image processing

Abstract

The present invention discloses a face feature recognition system based on digital image processing, which relates to the technical field of feature recognition, and includes a flaw recognition terminal, a retouching warning terminal, and an anomaly correction terminal; the flaw recognition terminal is used for determining the facial flaws of a human face in real time during multi-region recognition; the retouching warning terminal is used for adjusting the facial proportion of the five sense organs of the image in real time according to the standard proportion of the five sense organs, and judging the anomaly of the five sense organs proportion adjustment in real time in multiple directions, multiple angles and multiple positions; the anomaly correction terminal is used for calculating the retouching correction value in real time in all multiple directions, multiple angles and multiple positions. This face feature recognition system based on digital image processing can avoid the ineffective recognition of the spots and flaws on the faces of twins during face recognition, can judge whether the retouching is abnormal in real time according to the adjusted proportion after retouching, avoid the local deviation of the retouching after face feature recognition, and avoid the problem that the deviation of manual retouching affects the overall facial features.

Description

Technical Field

[0001] The present invention relates to the technical field of feature recognition, and particularly relates to a face feature recognition system based on digital image processing. Background Art

[0002] Face feature recognition based on digital image processing is a technology that uses digital image processing technology to analyze and identify feature information in face images. The color image is converted into a grayscale image to reduce the amount of data while highlighting key information such as the contour and texture of the image, facilitating subsequent processing. By adjusting parameters such as the contrast and brightness of the image, the face features in the image become more obvious, facilitating subsequent feature extraction.

[0003] Publication No. CN112651301A discloses an expression recognition method that integrates global and local face features. In the model feature extraction stage, three branches are designed; Branch One uses the entire face image to extract global face features; Branch Two cuts the face into two parts from top to bottom, and Branch Three cuts the face into three parts from top to bottom to extract local face features; the extracted global and local features are fused and then input into a softmax classifier for face expression classification, obtaining a recognition model that integrates global and local face features. Experiments show that the method of separately extracting and then fusing global and local features can improve the accuracy of face expression recognition.

[0004] After retrieving the above patents, it is found that there are still some deficiencies in face feature recognition based on digital image processing: 1. When performing face feature recognition, the spots and flaws on the face cannot be accurately recognized, resulting in the easy omission of the spots and flaws on the face during twin face recognition, leading to deviations in face feature recognition and affecting the accuracy of face feature recognition; 2. When performing retouching processing on the image obtained by face feature recognition, it is impossible to judge in real time whether the retouching is abnormal according to the adjustment ratio after retouching, resulting in easy local deviation of the retouching after face feature recognition. When there is local deviation, it cannot be reminded in time, affecting the effect of image processing; 3. When performing face feature recognition, the abnormal retouching usually depends on manual observation of whether the retouching deviates, and it is impossible to perform overall correction of the face according to local deviation, resulting in a significant reduction in the efficiency of image processing and high complexity of manual retouching.

[0005] Therefore, a face feature recognition system based on digital image processing is proposed to solve the above problems. Summary of the Invention

[0006] The main purpose of the present invention is to provide a face feature recognition system based on digital image processing to solve the problems raised in the above background.

[0007] To achieve the above object, the technical solution adopted by the present invention is as follows: A face feature recognition system based on digital image processing, comprising a defect recognition end, a photo retouching warning end, and an anomaly correction end;

[0008] The defect recognition end is used to randomly switch between local scanning and full scanning of the human face, perform multi-region recognition of human face features in real time, determine the positions of the facial features of the human face, and determine the defects of the human face in real time during multi-region recognition;

[0009] The photo retouching warning end is used to automatically generate a human face image based on the recognition result of the human face features, adjust the facial proportion of the five sense organs of the image in real time according to the standard proportion of the five sense organs, and judge in real time whether the adjustment of the local features affects all features, perform anomaly judgment on the adjustment of the proportion of the five sense organs in multiple directions, multiple angles, and multiple positions in real time, and when it is judged that the local adjustment of the proportion of the five sense organs affects all features, perform a photo retouching deviation warning reminder in real time;

[0010] The anomaly correction end is used to receive the local photo retouching anomaly judgment result of the image processing in real time through a data receiver, and perform real-time calculation of the photo retouching correction value in all directions, multiple angles, and multiple positions in real time when there is a local deviation in photo retouching, so that when there is a local deviation in image processing, all corrections and adjustments can be performed in real time, and perform photo retouching evaluation in real time after the image is corrected.

[0011] The defect recognition end includes a local switching module, a multi-effect recognition module, and a recognition classification module;

[0012] The local switching module includes a local scanning unit, a full scanning unit, and a five sense organs recognition and positioning unit;

[0013] The local scanning unit is used to collect the local features of the human face in real time through a 3D face scanner. The local features include eye color, eye texture, mouth color, mouth thickness, eyebrow shape, eyebrow length, eyebrow width, cheek color, cheek shape, philtrum length, philtrum width, spots, scars, acne marks, spot color, scar size, scar color, acne mark size, acne mark color, ear color, and ear shape, and record the local features of the human face in real time through a data recorder;

[0014] The full scanning unit is used to collect all the features of the human face in real time through a 3D face scanner. The all features include nose, eyes, mouth, eyebrows, cheeks, philtrum, and ears, and record all the features of the human face in real time through a data recorder;

[0015] The five sense organs recognition and positioning unit is used to set the five sense organs model of the human face features, determine the real-time positions of the five sense organs of the human face features in real time according to the scanning result of the 3D scanner, and display the five sense organs in real time through a local display page.

[0016] The multi-effect recognition module is used to determine in real time the facial defects of a human face during multi-region recognition. The facial defects include spots, spot colors, scars, and acne marks. It calculates in real time the color features of the human face and takes the average value of the facial defects , the average value of saturation and the average value of lightness as the color features of the facial defects. The calculation formula is as follows:

[0017] ;

[0018] ;

[0019] ;

[0020] where H represents hue, S represents saturation, V represents lightness, and represents the color feature component of the facial defect. , and are respectively the hue, saturation, and lightness values of the th pixel;

[0021] Set the standard color features of the facial features, and calculate the difference between the facial defect color and the standard color features. The calculation formula is as follows:

[0022] ;

[0023] where D represents the difference between the facial defect color and the standard color features. , and are respectively the average values of the hue, saturation, and lightness of the standard color. Identify the color at the positions where D is not equal to 0, indicating that there are facial defects here.

[0024] The recognition and classification module includes a defect classification unit, a feature identification unit, and a feature size unit;

[0025] The defect classification unit is used to classify the defects in real time according to the feature recognition results of the facial defects and in combination with the defect features of the 3D scanner. That is, the circular and swollen facial defects are recorded as acne marks, the irregularly shaped ones are recorded as scar facial defects, and the circular brown and circular black facial defects are recorded as spots;

[0026] The feature identification unit is used to identify different colors of spots with different colors through a color identifier;

[0027] The feature size unit is used to calculate in real time the size of the facial defects. The calculation formula is as follows:

[0028] The area of the circular spot is: A = N × L × w;

[0029] Wherein, N represents the number of pixels of the circular spot in the image, w represents the actual width corresponding to each pixel, and L represents the actual length corresponding to each pixel.

[0030] The image retouching warning terminal includes an image generation module, a retouching adjustment module, an abnormality judgment module, and a deviation warning module;

[0031] The image generation module is used to automatically generate a facial image of the facial features scanned by the 3D scanner through an image converter;

[0032] The retouching adjustment module includes a standard facial feature proportion unit and a facial feature proportion adjustment unit;

[0033] The standard facial feature proportion unit is used to set standard facial feature proportion parameters, that is, standard facial feature proportion parameters obtained according to different heights and weights, to obtain a standard facial feature proportion image;

[0034] The facial feature proportion adjustment unit is used to perform real-time adjustment of the facial feature proportion between the automatically generated facial image and the standard facial feature proportion image, and different-sized styluses are used for the adjustment.

[0035] The abnormality judgment module is used to combine multiple directions, multiple angles, and multiple positions to perform real-time abnormality judgment on the adjustment of facial feature proportion, specifically as follows:

[0036] Step 1, calculate all feature change values in multiple directions, multiple angles, and multiple positions , and the calculation formula is as follows:

[0037] ;

[0038] Wherein, represents all feature change values, represents the th predefined global proportion feature, represents the set of local feature points to be adjusted, represents the adjustment amount of the local feature points, represents the sensitivity of the global proportion to the local feature, and respectively represent the coordinate changes of the local point , and respectively represent the coordinates of the local point ;

[0039] Step 2, if is not equal to 0, it means that the facial feature proportion is abnormal and needs to be adjusted. If is equal to 0, it means that the facial feature proportion is normal and does not need to be adjusted.

[0040] The deviation warning module includes a retouching deviation warning unit, which is used to calculate the retouching deviation value in real time. The calculation method is as follows:

[0041] Set two points on the facial features as P1(x1, y1, z1) and P2(x2, y2, z2) respectively, and the two points are switched in turn according to the nose, mouth, eyebrows, ears and eyes to calculate the retouching deviation value The calculation formula is as follows:

[0042] ;

[0043] Set a deviation threshold. If the absolute value of is greater than the deviation threshold, it means that there is a deviation in retouching, and the reporting system issues a retouching anomaly warning reminder. Otherwise, it means that the retouching is normal.

[0044] The anomaly correction end includes a data receiving module, a global calculation module, a correction synchronization module and a correction evaluation module;

[0045] The data receiving module is used to receive the local retouching anomaly judgment result of image processing in real time through a data receiver.

[0046] The global calculation module includes a local deviation influence unit and a global calculation adjustment unit;

[0047] The local deviation influence unit is used to calculate the retouching correction value of the deviation in a progressive manner according to the distance of the facial features from small to large. The calculation method of the retouching correction value is as follows:

[0048] S1: Calculate the first-order coefficient of the regression equation. The formula is as follows:

[0049] ;

[0050] Among them, is the first-order coefficient, and it represents the quadratic relationship between the pitch angle value of the first retouching point and the correction angle data, specifically as follows: represents the basic value of the deviation of the first correction angle, represents the basic value of the deviation of the first correction angle reaching the specified point, represents the basic value of the deviation of the second correction angle, represents the basic value of the deviation of the second correction angle reaching the specified point, represents the basic value of the deviation of the correction angle at the current moment. Here, it represents the deviation values at different time points and different correction angles, and the correction angle represents the pitch correction angle of the facial features and the correction angle between the corresponding correction points;

[0051] S2: Calculate the quadratic term coefficient of the regression equation, and the formula is as follows:

[0052] ;

[0053] Among them, represents the quadratic term coefficient, and represents the binary quadratic relationship formed between the pitch angle value of the first retouching point and the correction angle data;

[0054] ;

[0055] Among them, represents the constant term coefficient, represents the average value of the correction angle deviation at the current moment. Here, the correction angle represents the corrected pitch angle and the correction angle between the specified correction point, represents the th deviation value of the correction angle, represents the average value of all deformation data in the pitch angle dataset during retouching point correction, represents the average value of all correction angle data in the correction angle dataset, represents the th deviation value of the correction angle of the correction point, that is, the sum of the deviation values of the correction angles between one correction point and several points of the facial features;

[0056] S3: According to the linear term coefficient, quadratic term coefficient and constant term coefficient, establish a regression equation, so as to obtain the correction deviation value of the pitch angle between the image correction and multiple facial features, and set the correction safety parameter;

[0057] The global calculation and adjustment unit is used to adjust the retouching points of the global features in real time according to the progressive retouching correction value calculated by the local deviation influence unit.

[0058] The correction synchronization module is used to track the progressive retouching correction value in real time through a data tracker, and perform retouching adjustment synchronously according to the progressive retouching correction value;

[0059] The correction evaluation module is used to perform image processing evaluation in real time after the retouching deviation is corrected. The evaluation method is to calculate the difference between the facial feature proportion parameter after the deviation is corrected and the standard facial feature proportion parameter. If the difference after equal-proportion magnification or reduction is less than or equal to 0.02 mm, it means that the facial feature proportion after image processing is normal. If the difference after equal-proportion magnification or reduction is greater than 0.02 mm, it means that the facial feature proportion after image processing is abnormal, and when the facial feature proportion after retouching is abnormal, it reports to the system and automatically displays the abnormal position through page display.

[0060] The present invention has the following beneficial effects:

[0061] 1. In the present invention, by setting up a defect recognition terminal, during the face feature recognition operation based on image processing, through the random switching between local scanning and full scanning of the human face, the multi-region recognition of the face features is carried out in real time, the positions of the facial features of the human face are determined in real time, and the defects of the human face are determined in real time during the multi-region recognition. The facial defects include spots, spot colors, scars, acne marks and lines, and the facial defects are classified in real time, the positions of the defects are marked with features, and the feature sizes are calculated in real time, so that when the system performs face feature recognition, it can accurately recognize the spot defects on the human face, avoid the ineffective recognition of the spot defects on the faces of twins during face recognition, prevent the deviation of face feature recognition, and further reduce the face feature recognition error.

[0062] 2. In the present invention, by setting up a retouching warning terminal, during the face feature recognition operation based on image processing, through the recognition result of the face features, a face image is automatically generated, the facial proportion of the five sense organs of the image is adjusted in real time according to the standard proportion of the five sense organs, and it is judged in real time whether the adjustment of the local features affects all the features. The abnormal judgment of the adjustment of the five sense organs proportion is carried out in multiple directions, multiple angles and multiple positions in real time, and when it is judged that the local adjustment of the five sense organs proportion affects all the features, a retouching deviation warning reminder is carried out in real time, so that when the system performs retouching processing, it can judge in real time whether the retouching is abnormal according to the adjusted proportion after retouching, avoid the local deviation of retouching after face feature recognition, and can timely remind the retoucher of the abnormal retouching of the human facial features when there is a local deviation, improving the retouching convenience effect of image processing.

[0063] 3. In the present invention, by setting up an abnormal correction terminal, during the face feature recognition operation based on image processing, the local retouching abnormal judgment result of image processing is received in real time through a data receiver, and when there is a local deviation in retouching, the retouching correction value is calculated in real time in all directions, multiple angles and multiple positions in real time, so that when there is a local deviation in image processing, all corrections can be carried out in real time, preventing the local retouching from affecting the processing effect of the overall image, solving the problem that manual observation cannot accurately identify the retouching deviation, and carrying out retouching adjustment synchronously through global calculation, so that the system can perform overall correction of the human face according to the local deviation, improving the efficiency of image processing, and through real-time retouching evaluation after image correction, so that after image processing, it can clearly know the retouching feedback, avoiding the problem that manual retouching deviation affects the overall facial features, and reducing the complexity of manual retouching. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 It is a schematic diagram of the overall system architecture of a face feature recognition system based on digital image processing according to the present invention;

[0065] Figure 2Schematic diagram of the framework of the defect recognition end of a face feature recognition system based on digital image processing according to the present invention;

[0066] Figure 3 Schematic diagram of the framework of the image retouching warning end of a face feature recognition system based on digital image processing according to the present invention;

[0067] Figure 4 Schematic diagram of the framework of the anomaly correction end of a face feature recognition system based on digital image processing according to the present invention. Detailed implementation manners

[0068] To make the technical means, creative features, achieved purposes and effects of the present invention easy to understand, the present invention will be further described below in conjunction with specific implementation manners.

[0069] Embodiment 1

[0070] Please refer to Figures 1 to 2 as shown: A face feature recognition system based on digital image processing includes a defect recognition end, an image retouching warning end and an anomaly correction end;

[0071] The defect recognition end is used to randomly switch between local scanning and full scanning of the human face, perform multi-region recognition of human face features in real time, determine the positions of the facial features of the human face, and determine the defects of the human face in real time during multi-region recognition;

[0072] The image retouching warning end is used to automatically generate a human face image based on the recognition result of the human face features, adjust the facial proportion of the five sense organs of the image in real time according to the standard proportion of the five sense organs, and judge in real time whether the adjustment of the local features affects all features, perform anomaly judgment on the adjustment of the five sense organs proportion in multiple directions, multiple angles and multiple positions in real time, and when it is judged that the local adjustment of the five sense organs proportion affects all features, give a real-time warning reminder of image retouching deviation;

[0073] The anomaly correction end is used to receive the local image retouching anomaly judgment result of the image processing in real time through a data receiver, and calculate the real-time image retouching correction value in all directions, multiple angles and multiple positions in real time when there is a local deviation in image retouching, so that when there is a local deviation in image processing, all corrections can be made in real time, and perform image retouching evaluation in real time after the image is corrected.

[0074] The defect recognition end includes a local switching module, a multi-effect recognition module and a recognition classification module;

[0075] The local switching module includes a local scanning unit, a full scanning unit and a facial feature recognition and positioning unit;

[0076] The local scanning unit is used to collect the local features of the human face in real time through a 3D face scanner. The local features include eye color, eye texture, mouth color, mouth thickness, eyebrow shape, eyebrow length, eyebrow width, cheek color, cheek shape, philtrum length, philtrum width, spots, scars, acne marks, spot color, scar size, scar color, acne mark size, acne mark color, ear color, and ear shape, and record the local features of the human face in real time through a data recorder;

[0077] The full scanning unit is used to collect all the features of the human face in real time through a 3D face scanner. The all features include nose, eyes, mouth, eyebrows, cheeks, philtrum, and ears, and record all the features of the human face in real time through a data recorder;

[0078] The facial feature recognition and positioning unit is used to set the facial feature models of the five sense organs of the human face, determine the real-time positions of the facial features of the five sense organs according to the scanning results of the 3D scanner, and display the five sense organs in real time through a local display page.

[0079] The multi-effect recognition module is used to determine the facial flaws of the human face in real time during multi-region recognition. The facial flaws include spots, spot color, scars, and acne marks, calculate the color features of the human face in real time, and take the average value of the facial flaws , the average value of saturation and the average value of lightness as the color features of the facial flaws. The calculation formula is as follows:

[0080] ;

[0081] ;

[0082] ;

[0083] where, H represents hue, S represents saturation, V represents lightness, represents the color feature component of the facial flaw, , and are the hue, saturation, and lightness values of the th pixel respectively;

[0084] Set the standard color features of the facial features, and calculate the difference between the facial flaw color and the standard color features. The calculation formula is as follows:

[0085] ;

[0086] where, D represents the difference between the facial flaw color and the standard color features, , and They are the average values of the hue, saturation, and lightness of the standard colors respectively. The positions where D is not equal to 0 are color-coded to indicate the presence of facial blemishes here.

[0087] The recognition and classification module includes a blemish classification unit, a feature identification unit, and a feature size unit;

[0088] The blemish classification unit is used to classify blemishes in real time according to the recognition results of facial blemish features and in combination with the blemish features of the 3D scanner. That is, the round and swollen facial blemishes are recorded as acne marks, the irregularly shaped ones are recorded as scar facial blemishes, and the round brown and round black facial blemishes are recorded as spots;

[0089] The feature identification unit is used to identify different colors of spots with a color identifier;

[0090] The feature size unit is used to calculate the size of facial blemishes in real time. The calculation formula is as follows:

[0091] The area of a round spot is: A = N×L×w;

[0092] Where N represents the number of pixels of the round spot in the image, w represents the actual width corresponding to each pixel, and L represents the actual length corresponding to each pixel. By classifying facial blemishes in real time, identifying the features of the blemish positions, and calculating the feature sizes in real time, the system can accurately identify the spot blemishes on the human face during face feature recognition, avoid the ineffective recognition of the spot blemishes on the faces of twins during face recognition, prevent deviations in face feature recognition, and further reduce the face feature recognition error.

[0093] Embodiment 2

[0094] Please refer to Figure 3 As shown: Based on Embodiment 1, the photo retouching warning terminal includes an image generation module, a photo retouching adjustment module, an anomaly judgment module, and a deviation warning module;

[0095] The image generation module is used for the image converter to automatically generate a facial image from the facial features scanned by the 3D scanner;

[0096] The photo retouching adjustment module includes a standard facial feature proportion unit and a facial feature proportion adjustment unit;

[0097] The standard facial feature proportion unit is used to set the standard facial feature proportion parameters, that is, the standard facial feature proportion parameters obtained according to different heights and weights, to obtain a standard facial feature proportion image;

[0098] The facial feature proportion adjustment unit is used to perform real-time adjustment of the facial feature proportions between the automatically generated facial image and the standard facial feature proportion image, and different-sized styluses are used for the adjustment.

[0099] The abnormal judgment module is used to combine multiple directions, multiple angles and multiple positions to judge the abnormality of the adjustment of the facial features in real time, specifically as follows:

[0100] Step 1: Calculate all the characteristic change values in multiple directions, multiple angles and multiple positions , and the calculation formula is as follows:

[0101] ;

[0102] Among them, represents all the characteristic change values, represents the th predefined global proportion feature, represents the set of local feature points to be adjusted, represents the adjustment amount of the local feature points, represents the sensitivity of the global proportion to the local feature, and respectively represent the coordinate changes of the local point , and respectively represent the coordinates of the local point ;

[0103] Step 2: If is not equal to 0, it means that the facial feature proportion is abnormal and needs to be adjusted. If is equal to 0, it means that the facial feature proportion is normal and does not need to be adjusted.

[0104] The deviation warning module includes a retouching deviation warning unit, and the retouching deviation warning unit is used to calculate the retouching deviation value in real time. The calculation method is as follows:

[0105] Set every two points on the facial features as P1(x1, y1, z1) and P2(x2, y2, z2) respectively, and the two points are switched in turn according to the nose, mouth, eyebrows, ears and eyes to calculate the retouching deviation value , and the calculation formula of

[0106] is as follows:

[0107] Set the deviation threshold. If the absolute value of is greater than the deviation threshold, it means that there is a deviation in retouching, and the system reports an abnormal warning reminder for retouching. Otherwise, it means that the retouching is normal. The abnormal judgment of the adjustment of the facial features in multiple directions, multiple angles and multiple positions is carried out in real time, and when it is judged that the local adjustment of the facial feature proportion affects all the features, a retouching deviation warning reminder is carried out in real time, so that when the system performs retouching processing, it can judge whether the retouching is abnormal in real time according to the adjusted proportion after retouching, and avoid local deviation of retouching after face feature recognition.

[0108] Embodiment III

[0109] Please refer to Figure 4 shown below: Based on the first embodiment, the anomaly correction end includes a data receiving module, a global calculation module, a correction synchronization module, and a correction evaluation module;

[0110] The data receiving module is used to receive the local retouching anomaly judgment result of image processing in real time through a data receiver.

[0111] The global calculation module includes a local deviation influence unit and a global calculation adjustment unit;

[0112] The local deviation influence unit is used to calculate the retouching correction value of the deviation in a progressive manner according to the distance of the facial features from small to large, or to calculate the edge deviation correction of the facial feature edges according to the following retouching correction value, so as to improve the accuracy of facial feature recognition, accurate to the angular deviation point-to-point. The retouching correction value calculation method is as follows:

[0113] S1: Calculate the first-order coefficient of the regression equation. The formula is as follows:

[0114] ;

[0115] Among them, is the first-order coefficient, and represents the unary quadratic relationship between the pitch angle value of the first retouching point and the correction angle data. Specifically, it is as follows: represents the basic value of the deviation of the first correction angle, represents the basic value of the deviation of the first correction angle reaching the specified point, represents the basic value of the deviation of the second correction angle, represents the basic value of the deviation of the second correction angle reaching the specified point, represents the basic value of the deviation of the correction angle at the current moment. Here, it represents the deviation values at different time points and different correction angles, and the correction angle represents the pitch correction angle of the facial features and the correction angle between the corresponding correction points;

[0116] S2: Calculate the second-order coefficient of the regression equation. The formula is as follows:

[0117] ;

[0118] Among them, represents the second-order coefficient, and represents the binary quadratic relationship between the pitch angle value of the first retouching point and the correction angle data;

[0119] ;

[0120] Among them, represents the constant term coefficient, represents the average value of the corrected angle deviation at the current moment. Here, the corrected angle refers to the corrected pitch angle and the corrected angle between the specified correction point, represents the deviation value of the th corrected angle, represents the average value of all deformation data in the pitch angle dataset when the retouching point is corrected, represents the average value of all corrected angle data in the corrected angle dataset, represents the deviation value of the corrected angle of the

[0121] S3: According to the linear term coefficient, quadratic term coefficient, and constant term coefficient, establish a regression equation, so as to obtain the corrected deviation value of the pitch angle between multiple facial features during image correction, and set the correction safety parameter;

[0122] The global calculation and adjustment unit is used to adjust the retouching points of global features in real time according to the progressive retouching correction value calculated by the local deviation influence unit. The adjustment of face data involved here is a legal adjustment, which is designed to avoid distortion still occurring after facial retouching.

[0123] The correction synchronization module is used to track the progressive retouching correction value in real time through a data tracker, and perform retouching adjustment synchronously according to the progressive retouching correction value;

[0124] The correction evaluation module is used to perform image processing evaluation in real time after the retouching deviation is corrected. The evaluation method is to calculate the difference between the proportion parameters of the facial features after deviation correction and the standard proportion parameters of the facial features. If the difference after equal-proportion magnification or reduction is less than or equal to 0.02 mm, it means that the proportion of the facial features after image processing is normal. If the difference after equal-proportion magnification or reduction is greater than 0.02 mm, it means that the proportion of the facial features after image processing is abnormal, and when the proportion of the facial features after retouching is abnormal, it reports to the system and automatically displays the abnormal position through page display, preventing local retouching from affecting the processing effect of the overall image, solving the problem that manual observation cannot accurately identify retouching deviation, and performing retouching adjustment synchronously through global calculation, enabling the system to perform overall correction of the human face according to local deviation, improving the efficiency of image processing, and through performing retouching evaluation in real time after image correction, enabling the system to clearly know the retouching feedback after image processing. The collection and use of human face data involved in the present invention are all legal collection and use.

[0125] In the present invention, a face feature recognition system based on digital image processing is provided. When the system operates, it can freely switch between local scanning and full scanning of the human face, and perform multi-region recognition of human face features in real time, determine the positions of the facial features of the human face in real time, and determine the facial defects of the human face in real time during multi-region recognition. Facial defects include spots, spot colors, scars, acne marks, and wrinkles, and classify the facial defects in real time, identify the features of the defect positions, and calculate the feature sizes in real time. When the system performs face feature recognition, it can accurately identify the spot defects on the human face, avoid the ineffective recognition of the spot defects on the faces of twins during face recognition, prevent deviation in face feature recognition, and further reduce the error of face feature recognition. An image of the human face is automatically generated based on the recognition result of the human face features, the facial proportion of the five sense organs of the image is adjusted in real time according to the standard proportion of the five sense organs, and it is judged in real time whether the adjustment of local features affects all features. Anomaly judgment of the adjustment of the five sense organs proportion is performed in multiple directions, multiple angles, and multiple positions in real time. When it is judged that the local adjustment of the five sense organs proportion affects all features, a warning reminder of retouching deviation is given in real time. When the system performs retouching processing, it can judge whether the retouching is abnormal in real time according to the adjusted proportion after retouching, avoid local deviation of the retouching after face feature recognition, and can timely remind the retoucher of the abnormality of the human face feature retouching when there is local deviation, improving the convenience effect of retouching in image processing. The local retouching anomaly judgment result of image processing is received in real time through a data receiver, and when there is local deviation in retouching, the retouching correction value is calculated in real time in all directions, multiple angles, and multiple positions. When there is local deviation in image processing, all corrections and adjustments can be performed in real time, preventing the local retouching from affecting the processing effect of the overall image, solving the problem that manual observation cannot accurately identify the retouching deviation, and performing retouching adjustment through global calculation synchronously, enabling the system to perform overall correction of the human face according to the local deviation, improving the efficiency of image processing, and performing retouching evaluation in real time after image correction, enabling the system to clearly know the retouching feedback after image processing, avoiding the problem that manual retouching deviation affects the overall facial features, and reducing the complexity of manual retouching.

[0126] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and the descriptions in the specification only illustrate the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A facial feature recognition system based on digital image processing, characterized in that: The facial feature recognition system based on digital image processing includes a defect recognition terminal, a photo editing warning terminal and an abnormality correction terminal; The defect recognition end is used to perform multi-region recognition of facial features of a person's face in real time by switching between partial scanning and full scanning of the face at will, determine the positions of the facial features of the person's face, and determine facial defects of the person's face in real time during multi-region recognition; The image editing warning terminal is used to automatically generate a face image of a human face based on the recognition result of the facial features of the human face, adjust the facial features of the image in real time according to the standard facial features ratio, and judge in real time whether the adjustment of local features affects all features, judge abnormalities of the facial features ratio adjustment in real time from multiple directions, angles and positions, and give a real-time image editing deviation warning reminder when it is judged that the local adjustment of the facial features ratio affects all features; The abnormality correction end is used to receive the local image retouching abnormality judgment result of the image processing in real time through the data receiver, and to perform all the real-time calculations of the image retouching correction values ​​in multiple directions, multiple angles and multiple positions in real time when the image retouching has local deviations, so that all correction adjustments can be performed in real time when the image processing has local deviations, and to perform the image retouching evaluation in real time after the image is corrected; The abnormal correction end includes a data receiving module, a global calculation module, a correction synchronization module and a correction evaluation module; The data receiving module is used to receive the local image repair abnormality judgment result of the image processing in real time through the data receiver; The global calculation module includes a local deviation influencing unit and a global calculation adjustment unit; The local deviation influence unit is used to calculate the correction value of the image retouching by progressively deviating the facial features in the order of distance from small to large. The calculation method of the image retouching correction value is as follows: S1: Calculate the coefficient of the first-order term of the regression equation. The formula is as follows: ； in, is the linear coefficient and represents the quadratic relationship between the pitch angle value of the first editing point and the correction angle data, as follows: Indicates the base value of the first correction angle deviation. Indicates the basic value of the deviation of the first correction angle to the specified point. Indicates the base value of the second correction angle deviation. Indicates the base value of the second correction angle to reach the specified point. Indicates the basic value of the correction angle deviation at the current moment. Here, it indicates the deviation values ​​at different time points and different correction angles. The correction angle indicates the pitch correction angle of the facial features and the correction angle between the facial features and the corresponding correction point. S2: Calculate the quadratic coefficient of the regression equation. The formula is as follows: ； in, represents the coefficient of the quadratic term, and represents the binary quadratic relationship between the pitch angle value of the first editing point and the correction angle data; ； in, represents the constant term coefficient, Indicates the average value of the correction angle deviation at the current moment. The correction angle here indicates the corrected pitch angle and the correction angle between the corrected angle and the specified correction point. Indicates The deviation value of the correction angle, It represents the average value of all deformation data in the pitch angle data set when the image editing point is corrected. represents the average value of all corrected angle data in the corrected angle data set. Indicates The deviation value of the correction angle of the correction point; S3: establishing a regression equation based on the linear term coefficient, the quadratic term coefficient and the constant term coefficient, thereby obtaining the correction deviation value of the pitch angle between the multiple facial features during image correction, and setting the correction safety parameter; The global calculation adjustment unit is used to adjust the photo retouching points of the global features in real time according to the progressive photo retouching correction value calculated by the local deviation influencing unit; The correction synchronization module is used to track the progressive image correction value in real time through a data tracker, and synchronously perform image correction adjustment according to the progressive image correction value; The correction evaluation module is used to perform image processing evaluation in real time after the deviation correction of the retouched image. The evaluation method is to calculate the difference between the facial feature proportion parameters after the deviation correction and the standard facial feature proportion parameters. If the difference after proportional enlargement or reduction is less than or equal to 0.02mm, it means that the proportion of facial features after image processing is normal. If the difference after proportional enlargement or reduction is greater than 0.02mm, it means that the proportion of facial features after image processing is abnormal. When the proportion of facial features after retouching is abnormal, it will be reported to the system and the abnormal position will be automatically displayed through the page display.

2. The system according to claim 1, characterized in that The defect recognition terminal includes a local switching module, a multi-effect recognition module and a recognition classification module; The local switching module includes a local scanning unit, a full scanning unit and a facial features recognition and positioning unit; The local scanning unit is used to collect local features of the human face in real time through a 3D facial scanner, the local features including eyeball color, eyeball texture, mouth color, mouth thickness, eyebrow shape, eyebrow length, eyebrow width, cheek color, cheek shape, philtrum length, philtrum width, spots, scars, acne marks, spot color, scar size, scar color, acne mark size, acne mark color, ear color and ear shape, and record the local features of the human face in real time through a data recorder; The full scanning unit is used to collect all features of the human face in real time through a 3D facial scanner, including nose, eyes, mouth, eyebrows, cheeks, philtrum and ears, and record all features of the human face in real time through a data recorder; The facial features recognition and positioning unit is used to set the facial features model of the human face, determine the real-time positions of the facial features of the human face in real time according to the scanning results of the 3D scanner, and display the facial features in real time through a local display page.

3. The system according to claim 2, characterized in that The multi-effect recognition module is used to determine the facial blemishes of the human face in real time during multi-region recognition. Facial blemishes include spots, spot colors, scars and acne marks. The color features of the human face are calculated in real time, and the average value of the facial blemishes is calculated. , the average value of saturation and the average value of brightness As the color feature of the facial blemish, the calculation formula is as follows: ； ； ； Among them, H represents hue, S represents saturation, and V represents brightness, which represents the color characteristic component of facial blemishes. , and They are The hue, saturation, and lightness values ​​of each pixel; Set the standard color features of facial features and calculate the difference between the color of facial blemishes and the standard color features. The calculation formula is as follows: ； Where D represents the difference between the color of facial blemishes and the standard color feature. , and They are the average values ​​of hue, saturation and brightness of the standard color respectively. The position where D is not equal to 0 is marked with color, indicating that facial blemishes appear here.

4. The system according to claim 3, characterized in that The identification and classification module includes a defect classification unit, a feature identification unit and a feature size unit; The defect classification unit is used to classify the defects in real time according to the feature recognition results of the facial defects and in combination with the defect features of the 3D scanner, that is, the round red and swollen facial defects are recorded as acne marks, the irregular shapes are recorded as scar facial defects, and the round brown and round black facial defects are recorded as spots; The characteristic identification unit is used to identify spots of different colors with different colors by using a color identification instrument; The characteristic size unit is used to calculate the size of facial blemishes in real time, and the calculation formula is as follows: The area of ​​a circular spot is: A = N × L × w; Among them, N represents the number of pixels of the circular spot in the image, w represents the actual width corresponding to each pixel, and L represents the actual length corresponding to each pixel.

5. The system according to claim 1, characterized in that The image editing warning terminal includes an image generation module, an image editing adjustment module, an abnormality judgment module and a deviation warning module; The image generation module is used for the image converter to automatically generate a facial image from the facial features of the human face scanned by the 3D scanner; The image editing and adjustment module includes a facial features standard ratio unit and a facial features ratio adjustment unit; The facial features standard proportion unit is used to set standard facial features proportion parameters, that is, standard facial features proportion parameters obtained according to different heights and weights, to obtain a standard facial features proportion image; The facial features ratio adjustment unit is used to adjust the facial features ratio of the automatically generated facial image and the standard facial features ratio image in real time, and the adjustment is performed using touch pens of different sizes.

6. The system according to claim 5, characterized in that The abnormality judgment module is used to judge abnormalities of facial features proportion adjustment in real time by combining multiple directions, multiple angles and multiple positions, as follows: Step 1: Calculate all feature change values ​​in multiple directions, angles and positions , the calculation formula is as follows: ； in, represents the total feature change value, Indicates predefined global scale features, represents the set of adjusted local feature points, Represents the adjustment amount of local feature points, represents the sensitivity of global scale to local features, and Represent local points The coordinate change of and Represent local points The coordinates of Step 2: If If it is not equal to 0, it means that the proportions of the facial features are abnormal and need to be adjusted. If it is equal to 0, it means that the proportions of facial features are normal and do not need to be adjusted.

7. The system according to claim 6, characterized in that The deviation warning module includes a modification deviation warning unit, which is used to calculate the modification deviation value in real time. The calculation method is as follows: Set every two points on the facial features as P1 (x1, y1, z1) and P2 (x2, y2, z2), and switch between the two points according to the nose, mouth, eyebrows, ears and eyes to calculate the deviation value of the retouching The calculation formula is as follows: ； Set the deviation threshold. If If the absolute value of is greater than the deviation threshold, it means that the image editing has deviated, and the reporting system issues an abnormal image editing warning reminder. If not, it means that the image editing is normal.

Images