On-site Fingerprint Search and Evidence Collection Device and Its Evidence Collection Method

Abstract

A device for on-site fingerprint search and evidence collection includes a mobile housing, an imager, a combined light source, a combined light source lifting mechanism, and a controller. The imager and the combined light source are respectively arranged inside the mobile housing, the imager is located above the combined light source, and the combined light source includes a mercury lamp and a multi-spectrum LED light source. The combined light source lifting mechanism and the controller are respectively fixed to the mobile housing, and the combined light source lifting mechanism is used to drive the combined light source to rise and fall. The signal input end of the controller is connected to the signal output end of the imager, the output end of the controller is connected to the control input end of the combined light source lifting mechanism, and the controller has a display screen to display the images collected by the imager. The present invention also discloses a method for on-site fingerprint search and evidence collection. The present invention can collect fingerprint images with multi-spectrum and multi-angle light distribution and is easy to operate.

Description

Technical Field

[0001] The present invention relates to the field of forensic science, and particularly to the evidence collection technology of on-site fingerprints. Background Art

[0002] In order to give full play to the application of various optical inspection methods in physical evidence investigation, foreign countries began to try to use multi-band light sources to replace lasers at the end of the 1980s. The multi-band light source generally consists of a light source system, a color filter system and a light guide tube. At present, a tungsten halogen lamp or a xenon lamp is generally used as the light source. The lights of the tungsten halogen lamp and the xenon lamp are relatively pure and have high efficiency, but they cannot be used without power supply and are bulky and not easy to carry. In addition, the filter and bulb life of this filtering multi-band light source are very limited and need to be replaced frequently. In order to solve a series of problems of the current filtering multi-band light source such as high power consumption, large heat generation and short life, some domestic and foreign manufacturers use light-emitting diodes (LEDs) as the light source and develop low-power multi-band light sources. In order to reduce the price and have the function of multiple bands at the same time, Luyuan LUYOR-3220P OLICE in the United States launched a multi-band flashlight / LED inspection lamp, providing a complete spectral light source. The eight flashlights respectively output ultraviolet, purple, blue, cyan, green, yellow, red and white light. However, for the specific emission wavelengths of certain substances, the complementary cooperation of multiple monochromatic light sources is required, which will cause certain difficulties in extracting effective physical evidence and is not conducive to use.

[0003] In recent years, the red-ultraviolet imaging technology has developed rapidly. In particular, the development of red-ultraviolet visualization technology has created good application conditions for the development of this technology. Special equipment such as red-ultraviolet observation and photography systems, full-band CCDs, and full-spectrum special photography evidence collection devices have been successively equipped in various places. In addition, Wang Zhendong et al. reported the application of red-ultraviolet imaging technology to reveal latent handwriting on hats and gloves; Huang Wei et al. reported the application of red-ultraviolet imaging technology to reveal latent red handwriting on sheets; Zhang Xinguo et al. reported the application of red-ultraviolet observation and photography systems to reveal sweat fingerprints; Wang Xufeng et al. reported the application of full-band CCD systems in ultraviolet reflection photography, etc. Abroad, research on using red-ultraviolet imaging technology to reveal latent physical evidence has been carried out since the 1960s. The American Spex Optics Company specifically produces the SceneScope red-ultraviolet observation and photography system for revealing latent fingerprints. The image intensifier of this system has a service life of more than 10,000 hours, high sensitivity, and a large signal-to-noise ratio. Combined with high-quality ultraviolet filter lenses, high-performance ultraviolet lenses, and military-grade eyepieces, the imaging is clearer, and it is more suitable for the discovery and extraction of faint sweat fingerprints. It integrates search, observation, and photography, and can achieve long-distance search and close-range photography; the VSC series of instruments developed by the British Foster Freeman Company integrates various bands such as infrared, ultraviolet, and visible light; the ultraviolet observation and photography system of the Dutch DEP Company enables the operator to directly observe visible fingerprint images through the image conversion function of a super-strong ultraviolet intensifying tube, and at the same time, the observed images can be directly recorded using film, digital cameras, etc.

[0004] In summary, certain achievements have been made in the optical revelation of latent physical evidence at home and abroad. However, there are a wide variety of optical revelation methods and the operations are relatively cumbersome. Differences in the technical levels of personnel often affect the quality of evidence collection. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a device for searching and collecting on-site fingerprint evidence, which can collect fingerprint images with multi-spectrum and multi-angle light distribution, improve the revelation rate of latent fingerprints, and has a compact structure, is easy to operate, and has a low implementation cost.

[0006] Another technical problem to be solved by the present invention is to provide a method for searching and collecting on-site fingerprint evidence.

[0007] According to one aspect of an embodiment of the present invention, a device for searching, collecting and obtaining on-site fingerprints is provided, including a mobile housing, an imager, a combined light source, a combined light source lifting mechanism and a controller; the imager and the combined light source are respectively arranged inside the mobile housing, the imager is located above the combined light source, and the combined light source includes a mercury lamp and a multi-spectral LED light source; the combined light source lifting mechanism and the controller are respectively fixed to the mobile housing, and the combined light source lifting mechanism is used to drive the combined light source to rise and fall; the signal input end of the controller is connected to the signal output end of the imager, the output end of the controller is connected to the control input end of the combined light source lifting mechanism, and the controller has a display screen to display the images collected by the imager.

[0008] According to another aspect of an embodiment of the present invention, a method for collecting evidence of the above-mentioned on-site fingerprint search and collection device is further provided, which is characterized by including the following steps:

[0009] S1. The on-site fingerprint search and collection device receives a power-on instruction, turns on the imager, and makes the display screen display real-time images;

[0010] S2. Align the mobile housing of the on-site fingerprint search and collection device to the on-site fingerprints found by moving it.

[0011] S3. Make the on-site fingerprint search and collection device first work in the reflection mode, and then work in the fluorescence mode;

[0012] When the on-site fingerprint search and collection device is in the reflection mode, the following steps are sequentially executed:

[0013] S31. The combined light source lifting mechanism drives the combined light source to move to a preset highest position, and then sequentially turns on the short-wave ultraviolet lamp, long-wave ultraviolet LED lamp, violet LED lamp, blue LED lamp, green LED lamp and near-infrared LED lamp, and respectively obtains 6 fingerprint photos by imaging;

[0014] S32. The combined light source lifting mechanism drives the combined light source to move to a preset intermediate position, and then sequentially turns on the short-wave ultraviolet lamp, long-wave ultraviolet LED lamp, violet LED lamp, blue LED lamp, green LED lamp and near-infrared LED lamp, and respectively obtains 6 fingerprint photos by imaging;

[0015] S33. The combined light source lifting mechanism drives the combined light source to move to a preset lowest position, and then sequentially turns on the short-wave ultraviolet lamp, long-wave ultraviolet LED lamp, violet LED lamp, blue LED lamp, green LED lamp and near-infrared LED lamp, and respectively obtains 6 fingerprint photos by imaging;

[0016] When the on-site fingerprint search and collection device is in the fluorescence mode, the following steps are sequentially executed:

[0017] S34. The filter switching mechanism moves the 550 nm long-pass filter between the lens of the imager and the CCD image sensor. The combined light source lifting mechanism drives the combined light source to move to a preset intermediate position. Then, the short-wave ultraviolet lamp, long-wave ultraviolet LED lamp, violet LED lamp, blue LED lamp, and green LED lamp are turned on in sequence, and 5 fingerprint photos are obtained by imaging respectively;

[0018] S35. The filter switching mechanism removes the 550 nm long-pass filter and moves the 580 nm long-pass filter between the lens of the imager and the CCD image sensor. Then, the short-wave ultraviolet lamp, long-wave ultraviolet LED lamp, violet LED lamp, blue LED lamp, and green LED lamp are turned on in sequence, and 5 fingerprint photos are obtained by imaging respectively;

[0019] S36. The filter switching mechanism removes the 580 nm long-pass filter and moves the 600 nm long-pass filter between the lens of the imager and the CCD image sensor. Then, the short-wave ultraviolet lamp, long-wave ultraviolet LED lamp, violet LED lamp, blue LED lamp, and green LED lamp are turned on in sequence, and 5 fingerprint photos are obtained by imaging respectively;

[0020] S4. The controller performs geometric registration, feature extraction, feature optimization, and feature fusion on the 33 fingerprint photos obtained in step S3 to obtain an optimized fingerprint photo, automatically saves the optimized fingerprint photo, and displays it on the display screen.

[0021] 1. The on-site fingerprint search and evidence collection device of the present invention is provided with a combined light source and a combined light source lifting mechanism, which can use light sources of multiple spectral bands to take multiple fingerprint images at multiple light distribution illumination angles, so as to comprehensively photograph the detailed features of the fingerprint, improve the clarity of the fingerprint image and the appearance rate of latent fingerprints;

[0022] 2. The on-site fingerprint search and evidence collection device of the embodiment of the present invention integrates the imager, combined light source, combined light source lifting mechanism, and controller into a mobile housing, with a compact structure, easy to operate, and low implementation cost;

[0023] 3. The on-site fingerprint search and evidence collection device according to the embodiment of the present invention can simplify the evidence collection steps, and can quickly complete the appearance and extraction of on-site latent traces with just one key operation, avoiding the influence of cumbersome operation steps and differences in personnel technical levels on the evidence collection quality, and meeting the needs of rapid disposal work at the criminal case scene;

[0024] 4. The on-site fingerprint search and evidence collection method according to the embodiment of the present invention performs feature fusion through non-downsampled morphological Haar wavelet transform, which can eliminate the block effect after coefficient processing, thereby effectively improving the image fusion effect. Brief Description of the Drawings

[0025] Figure 1 The structural schematic diagram of a on-site fingerprint search and evidence collection device according to an embodiment of the present invention is shown.

[0026] Figure 2 The control principle block diagram of a on-site fingerprint search and evidence collection device according to an embodiment of the present invention is shown.

[0027] Figure 3 The layout schematic diagram of an annular multi-spectrum LED light source according to an embodiment of the present invention is shown.

[0028] Figures 4 to 6 The definition of morphological operators, the equivalent translocation of morphological operators, and the equivalent translocation conversion of scale function sub-bands are respectively described.

[0029] Figure 7 The process example of an image fusion algorithm based on UMHWT is shown. Specific embodiments

[0030] A further description of the present invention will be made below with reference to the accompanying drawings.

[0031] Please refer to Figure 1 and Figure 2 . A on-site fingerprint search and evidence collection device according to an embodiment of the present invention includes a mobile housing 1, an imager 2, a combined light source 3, a combined light source lifting mechanism, and a controller 5.

[0032] The imager 2 and the combined light source 3 are respectively arranged inside the mobile housing 1. The imager 2 is located above the combined light source 3. The combined light source 3 includes a mercury lamp 31 and a multi-spectrum LED light source 32.

[0033] The combined light source lifting mechanism and the controller 5 are respectively fixed to the mobile housing 1. The combined light source lifting mechanism is used to drive the combined light source 3 to rise and fall. The signal input end of the controller 5 is connected to the signal output end of the imager 2, and the output end of the controller 5 is connected to the control input end of the combined light source lifting mechanism to control the lifting of the combined light source 3. The controller 5 has a display screen to display the images collected by the imager 2.

[0034] In this embodiment, the combined light source 3 has a light source base 30, and a mercury lamp 31 and a multi-spectrum LED light source 32 are respectively fixed on the light source base 30. The combined light source lifting mechanism includes a lifting stepper motor 41, a main gear 42, a pair of slave gears 43 and a pair of wire reels 44. The lifting stepper motor 41 is fixed to the moving housing 1. The output end of the lifting stepper motor 41 is connected to the main gear 42. A pair of slave gears 43 are respectively located on opposite sides of the main gear 42 and mesh with the main gear 42. A pair of wire reels 44 are coaxially arranged with a pair of slave gears 43 respectively (that is, the wire reel and the slave gear are installed on the same axle). Suspension wires 45 are wound around each wire reel, and the suspension wires 45 on the pair of wire reels are respectively connected to the light source base 30. By controlling the lifting stepper motor 41, the suspension wires 45 on the pair of wire reels can be driven to drive the combined light source 3 to move up and down, thereby changing the irradiation angle of the light source.

[0035] The mercury lamp 31 is an annular mercury lamp, and the annular mercury lamp is a 254nm short-wave ultraviolet lamp. The multi-spectrum LED light source 32 is an annular multi-spectrum LED light source. The annular mercury lamp is located on the periphery of the annular multi-spectrum LED light source. That is, the annular mercury lamp constitutes the outer ring light source, and the annular multi-spectrum LED light source constitutes the inner ring light source. The annular multi-spectrum LED light source is composed of a long-wave ultraviolet LED lamp, a violet LED lamp, a blue LED lamp, a green LED lamp and a near-infrared LED lamp.

[0036] Please refer to Figure 3 . In this embodiment, the long-wave ultraviolet LED lamp, the violet LED lamp, the blue LED lamp, the green LED lamp and the near-infrared LED lamp are respectively composed of 4 long-wave ultraviolet LED lamp beads 321, 4 violet LED lamp beads 322, 4 blue LED lamp beads 323, 4 green LED lamp beads 324 and 4 near-infrared LED lamp beads 325, and the long-wave ultraviolet LED lamp beads 321, the violet LED lamp beads 322, the blue LED lamp beads 323, the green LED lamp beads 324 and the near-infrared LED lamp beads 325 are arranged in an interleaved manner; among the 4 LED lamp beads of each color, the connection line between 2 of the LED lamp beads passes through the center of the circle, and the connection line between the remaining 2 LED lamp beads also passes through the center of the circle, and the connection line between 2 of the LED lamp beads is orthogonal (perpendicular) to the connection line between the remaining 2 LED lamp beads. This arrangement can make the illumination brightness distribution uniform.

[0037] In this embodiment, the controller 5 is a computer. For example, a tablet computer can be used. By acquiring the image signal output by the imager, the computer can perform image processing on the image signal sent by the imager and display the image after image processing. The image processing includes geometric registration, feature extraction, feature optimization and feature fusion.

[0038] In this embodiment, geometric registration is performed by using translation transformation, scaling transformation, and rotation transformation in sequence to make the images of the same area obtained with different bands and different light distribution angles completely overlap. Feature extraction is to extract the color features, texture features, shape features, and spatial relationship features of the geometrically registered images in sequence. Feature optimization is to enhance and optimize the image features through the recognition of fingerprint line features and in combination with image processing methods such as brightness and color level adjustment. Feature fusion is to fuse the useful feature information in multiple images and comprehensively display it in one image.

[0039] The imager 2 includes a lens 21, a filter 22, and a CCD image sensor 23 that are arranged in sequence from bottom to top. The signal output end of the CCD image sensor 23 is connected to the signal input end of the controller 5.

[0040] In this embodiment, the imager includes a 550nm long-pass filter, a 580nm long-pass filter, a 600nm long-pass filter, and a filter switching mechanism. The filter switching mechanism is used to selectively move one of the multiple filters to between the lens 21 and the CCD image sensor 23 under the control of the controller 5. In a specific implementation manner, the filter switching mechanism includes a lens wheel for placing multiple filters and a rotary stepping motor 24 for driving the rotation of the lens wheel. The controller 5 can control the rotation of the rotary stepping motor 24 to achieve filter switching. In other implementation manners, the filter can also be switched manually by plugging and unplugging.

[0041] In this embodiment, the on-site fingerprint search and evidence collection device further includes a light source drive circuit 6. The light source drive circuit is used to control the turning on and off of the combined light source under the control of the controller, so as to achieve the automatic control of the turning on and off of the 254nm short-wave ultraviolet lamp, long-wave ultraviolet LED lamp, violet LED lamp, blue LED lamp, green LED lamp, and near-infrared LED lamp. In other implementation manners, the turning on and off of the 254nm short-wave ultraviolet lamp, long-wave ultraviolet LED lamp, violet LED lamp, blue LED lamp, green LED lamp, and near-infrared LED lamp can also be controlled by a manual switch.

[0042] Further, the moving housing 1 includes an upper housing 1a and a lower housing 1b. The top of the lower housing 1b is connected to the bottom of the upper housing 1a. Multiple moving wheels 1c are provided at the bottom of the lower housing 1b. In this embodiment, 3 universal wheels are installed at the bottom of the lower housing 1b. The imager 3 and the controller 5 are respectively arranged in the upper housing 1a. Among them, the touch screen of the tablet computer serving as the controller 5 is arranged on the top of the upper housing 1a for the convenience of operation by the operator, and the combined light source 3 is arranged in the inner cavity of the lower housing 1b.

[0043] According to another aspect of the embodiments of the present invention, there is also provided a forensics method using the above-mentioned on-site fingerprint search and forensics device, including the following steps:

[0044] S1. The on-site fingerprint search and forensics device receives a power-on instruction, lights up all the LED lamp beads, and turns on the imager 2, so that the display screen of the controller 5 displays real-time images;

[0045] S2. By moving the moving housing 1 of the on-site fingerprint search and forensics device, align it with the searched on-site fingerprint; when the image of the on-site fingerprint is clearly displayed on the display screen, it can be considered that the searched on-site fingerprint has been aligned;

[0046] S3. Make the on-site fingerprint search and forensics device work in the reflection mode first, and then work in the fluorescence mode;

[0047] When the on-site fingerprint search and forensics device is in the reflection mode, the following steps are sequentially executed:

[0048] S31. The combined light source lifting mechanism drives the combined light source 3 to move to a preset highest position, and then sequentially turns on the short-wave ultraviolet lamp, long-wave ultraviolet LED lamp, violet LED lamp, blue LED lamp, green LED lamp, and near-infrared LED lamp, and respectively obtains 6 fingerprint photos by imaging; it should be noted that among the above 6 light sources, whenever one of the light sources is turned on, the remaining 5 light sources are turned off, and the same is true for the following steps S32 and S33;

[0049] S32. The combined light source lifting mechanism drives the combined light source to move to a preset intermediate position, and then sequentially turns on the short-wave ultraviolet lamp, long-wave ultraviolet LED lamp, violet LED lamp, blue LED lamp, green LED lamp, and near-infrared LED lamp, and respectively obtains 6 fingerprint photos by imaging;

[0050] S33. The combined light source lifting mechanism drives the combined light source to move to a preset lowest position, and then sequentially turns on the short-wave ultraviolet lamp, long-wave ultraviolet LED lamp, violet LED lamp, blue LED lamp, green LED lamp, and near-infrared LED lamp, and respectively obtains 6 fingerprint photos by imaging;

[0051] When the on-site fingerprint search and forensics device is in the fluorescence mode, the following steps are sequentially executed:

[0052] S34. The filter switching mechanism moves the 550 nm long - pass filter to between the lens 21 of the imager and the CCD image sensor 23. The combined light source lifting mechanism drives the combined light source to move to a preset intermediate position, and then the short - wave ultraviolet lamp, long - wave ultraviolet LED lamp, violet LED lamp, blue LED lamp, and green LED lamp are sequentially turned on, and 5 fingerprint photos are obtained by imaging respectively. It should be noted that among the above 5 light sources, whenever one of the light sources is turned on, the other 4 light sources are turned off, and the same is true for the following step S35 and step S36;

[0053] S35. The filter switching mechanism removes the 550 nm long - pass filter and moves the 580 nm long - pass filter to between the lens 21 of the imager and the CCD image sensor 23, and then the short - wave ultraviolet lamp, long - wave ultraviolet LED lamp, violet LED lamp, blue LED lamp, and green LED lamp are sequentially turned on, and 5 fingerprint photos are obtained by imaging respectively;

[0054] S36. The filter switching mechanism removes the 580 nm long - pass filter and moves the 600 nm long - pass filter to between the lens of the imager and the CCD image sensor, and then the short - wave ultraviolet lamp, long - wave ultraviolet LED lamp, violet LED lamp, blue LED lamp, and green LED lamp are sequentially turned on, and 5 fingerprint photos are obtained by imaging respectively;

[0055] S4. The controller performs geometric registration, feature extraction, feature optimization, and feature fusion on the 33 fingerprint photos obtained in steps S31 to S36 to obtain an optimized fingerprint photo, automatically saves the optimized fingerprint photo, and displays it on the screen.

[0056] Preferably, in this embodiment, in step S3, after receiving the one - key evidence - collection instruction, the on - site fingerprint search and evidence - collection device automatically works in the reflection mode first and then in the fluorescence mode. The user only needs to press one key to quickly complete the visualization and extraction of potential traces at the scene, avoiding the influence of cumbersome operation steps and differences in personnel technical levels on the quality of evidence collection, and meeting the needs of rapid disposal work at the criminal case scene.

[0057] In this embodiment, the feature fusion includes the following steps:

[0058] Perform undecimated morphological Haar wavelet transform (UMHWT) on the 33 fingerprint photos obtained in steps S31 to S36 respectively to obtain the low - frequency coefficients (i.e., decomposition coefficients) at the total scale number J (i.e., the decomposition layer number of UMHWT) and the high - frequency coefficients (i.e., scaling function coefficients) at each scale j (0 < j ≤ J);

[0059] Fuse the low - frequency coefficients according to the fusion rule of low - frequency coefficients, and fuse the high - frequency coefficients according to the fusion rule of high - frequency coefficients;

[0060] Perform consistency verification on the obtained low-frequency fusion coefficients and high-frequency fusion coefficients respectively, and update the low-frequency fusion coefficients and high-frequency fusion coefficients using the detection results;

[0061] Perform the inverse transform of the undecimated morphological Haar wavelet transform (Inverse UMHWT, IUMHWT) on the final low-frequency fusion coefficients and high-frequency fusion coefficients to obtain the final fused image.

[0062] The undecimated morphological Haar wavelet transform (UMHWT for short) is obtained by using an undecimated method to perform a shift-invariant extension on the morphological Haar wavelet transform to address the problem of poor quality of the fused image due to the lack of shift invariance in traditional wavelet transform-based image fusion algorithms.

[0063] For the implementation method of UMHWT in the one-dimensional case, first define the morphological filtering operator S(k) as Figure 4 shown. This operator satisfies the Noble Identities, and its equivalence transposition property can be represented by Figure 5 Then, the decomposition of the morphological Haar wavelet transform (MHWT) on the scaling function sub-band is iteratively implemented through the S(1) operator and the downsampling operator ↓2. Finally, the combined implementation form of this decomposition is obtained using the Noble Identities, as Figure 6 shown, which is the undecimated shift-invariant extension of the decomposition coefficients of the i-th level wavelet on the scaling function sub-band:

[0064]

[0065] Therefore, the decomposition of UMHWT on the i-th level wavelet scaling function sub-band can be iteratively expressed as:

[0066]

[0067] where, represents the original input signal. Similarly, the decomposition of UHMWT on the i-th level wavelet sub-band can be iteratively expressed as:

[0068]

[0069] Furthermore, the reconstruction of UHMWT can be iteratively expressed as:

[0070]

[0071]

[0072] The sum of the output components of each channel can still reconstruct the low-frequency coefficients of the previous level. Therefore, the morphological Haar wavelet transform after downsampling removal still has perfect reconstruction property.

[0073] In the above formula, represents the original input signal, where n represents the region or range corresponding to the input signal data (for example, if the input x is an image, then x(n) represents the pixel value of the input image region); "∨" and "∧" respectively represent the "dilation" operator and the "erosion" operator, that is, the maximum value or minimum value operation, k ∈ [0, 2 i ); and respectively represent the approximation coefficient (i.e., the decomposition coefficient on the scaling function sub-band) and the detail coefficient (i.e., the decomposition coefficient on the wavelet sub-band) of the i-th level wavelet decomposition of the morphological Haar wavelet transform (MHWT), is the decomposition coefficient is the shift-invariant extension without downsampling of the decomposition coefficient; Formulas (4) and (5) represent the reconstruction process of UMHWT, where and respectively represent the scaling function of the reconstructed i-th level wavelet and the output components of the wavelet (i.e., the reconstructed approximation coefficient and detail coefficient).

[0074] For the implementation method of UMHWT in the two-dimensional case, the case of extending the standard discrete wavelet transform to two dimensions can be referred to, and UMHWT is extended to two dimensions by adopting a row-column separable method. Similar to the one-dimensional case, the decomposition iteration expressions and reconstruction iteration expressions on the four channels of two-dimensional UMHWT can also be derived according to the transposition equivalence of the morphological operators. In addition, it is necessary to ensure that UMHWT still retains the advantageous characteristics of MHWT, and the scaling function coefficients at each level have the same value range as the input image signal. And if the input image signal is integer quantization, then the scaling function and wavelet decomposition coefficients are also integer quantization. This is very meaningful when processing image signals. For example, for a 256-level two-dimensional grayscale image, the decomposition coefficients at each level are integers, and the scaling function decomposition coefficients have the same value range as the input image, that is, [0, 255]. The scaling function decomposition coefficient is the maximum value of the relevant region of the input image, and the higher the level, the larger the area of the relevant region. The wavelet decomposition coefficient retains the detail information of the input image at different scales.

[0075] Applying the shift-invariant extension form of the non-linear Haar wavelet obtained above to the multi-scale image fusion scheme, the multi-focus image fusion algorithm based on UMHWT can be obtained.

[0076] Let the source images be I1 and I2 respectively, and the fused image be IF. The process of the image fusion algorithm based on UMHWT is as follows Figure 7 shown. The main steps are as follows:

[0077] First, perform UMHWT on the source images I1 and I2 respectively to obtain the low-frequency coefficients (i.e., decomposition coefficients) at the total scale number J (i.e., the decomposition layer number of UMHWT) and the high-frequency coefficients (i.e., scaling function coefficients) at each scale j (0 < j ≤ J); fuse the low-frequency coefficients according to the fusion rule of low-frequency coefficients, and fuse the high-frequency coefficients according to the fusion rule of high-frequency coefficients; perform consistency verification on the obtained low-frequency fusion coefficients and high-frequency fusion coefficients respectively, and use the detection results to update the low-frequency fusion coefficients and high-frequency fusion coefficients; perform the inverse transform of UMHWT (Inverse UMHWT, IUMHWT) on the final low-frequency fusion coefficients and high-frequency fusion coefficients to obtain the final fused image IF.

[0078] After obtaining the low-frequency approximation coefficients (i.e., the aforementioned low-frequency coefficients) and high-frequency detail coefficients (i.e., the aforementioned high-frequency coefficients) of the source image through UMHWT, a key step in image fusion is to combine these coefficients. Since the physical meanings of the coefficients are different, different fusion rules need to be used to process them separately. The low-frequency coefficients mainly reflect the gray information of the image and include most of the image energy. A relatively simple fusion rule is to average the low-frequency coefficients of the source image. However, when there are obvious differences between the source images, this rule may reduce the contrast of the fused image. The high-frequency coefficients mainly include detailed information such as the edges, contours, and textures of the image. Generally, the larger the absolute value of the coefficient, the more obvious the corresponding detailed information. Therefore, the fusion rule for high-frequency coefficients usually selects the coefficient with the larger absolute value at the same position as the fusion coefficient. Both of these rules operate on each coefficient individually and are generally referred to as the Coefficient Based Fusion Rule (CBFR). The coefficient point fusion rule ignores the correlation between coefficients, may result in an inappropriate coefficient mapping method, and is easily affected by noise. To improve the quality of the fused image, a more reasonable approach is to consider the coefficients in a certain window or region of the source image as a whole, namely the Window Based Fusion Rule and the Region Based Fusion Rule. The window fusion rule takes the coefficients at each point as the center of the window and jointly determines the fusion behavior of the current coefficient through the characteristic information of the coefficients in the window. After this step is completed, the window center is moved to the next adjacent coefficient, and the same process is carried out. The region fusion rule divides the image into regions for fusion. Therefore, an image segmentation algorithm is first required to divide the image into regions, and the region fusion process is determined according to the characteristic information corresponding to each region. Since this rule involves an image segmentation algorithm, it will inevitably increase the computational complexity and implementation difficulty, so it is not conducive to real-time processing, and the fusion effect largely depends on the accuracy of the segmentation algorithm.

[0079] In this embodiment, the window fusion rule is adopted, which is more suitable for most image processing tasks including forensic image processing. For multi-focus image fusion, since it is desired to extract the best-focused part of the source image and then splice it into a fused image. Therefore, the source of the fusion coefficient at each position should be single, rather than a combination of source coefficients.

[0080] The low-frequency coefficients are the approximate images obtained by low-pass filtering the source image. The focus evaluation function can be used to judge its focus degree. Moreover, the noise of the source image mainly exists in the high-frequency coefficients and will not affect the evaluation result. Therefore, in this embodiment, the focus evaluation function is used to guide the fusion of the low-frequency coefficients. Since the focus evaluation function generally considers the entire image, and in the fusion process, it is necessary to judge the focus of a certain part of the image. Therefore, when using the window fusion rule, the coefficients covered by the window are regarded as an independent image. At this time, the focus evaluation function only judges the focus degree of the coefficients in the window. According to the definition of the window fusion rule, the focus evaluation function value of the coefficients in the window is used as the basis for the fusion operation of the current coefficient. The source image is subjected to UMHWT to obtain high-frequency coefficients of multiple scales and directions. Each high-frequency coefficient has the same size as the low-frequency coefficient, and the coefficients at the same position correspond to the same physical space. Therefore, the high-frequency coefficients can use the same mapping method as the low-frequency coefficients. In addition, for multi-focus image fusion, if the sources of the high-frequency fusion coefficients and the low-frequency fusion coefficients are the same, the information of the source image can be presented to the greatest extent, and some "edge effects" can be avoided. However, in general, it is impossible to ensure the accuracy of the mapping method of the low-frequency coefficients. If the same mapping method is used for the high-frequency coefficients, it may cause the loss of some high-frequency information. For most image processing tasks, the high-frequency information of the image is more important than the low-frequency information. Similarly, if the mapping method of the high-frequency coefficients is used to indicate the fusion of the low-frequency coefficients, similar problems will also occur. If the coefficients of each scale are fused independently without considering their spatial relationship, it is possible that the sources of the fused coefficients of each scale at the same position are inconsistent. This inconsistency will cause changes in the information of the source image in the fused image, but at least can partially maintain the gray information or detail information of the image. Compared with the consistent fusion method under inaccurate mapping, the independent coefficient fusion method can retain more information of the source image. To sum up, in order to ensure the stability of the fusion algorithm, the fusion rule set for the high-frequency coefficients in this embodiment does not need to refer to the fusion method of the low-frequency coefficients, nor does it need to ensure that the coefficients of each scale have the same mapping method.

[0081] Feature fusion can be performed by non-subsampled morphological Haar wavelet transform to eliminate the block effect after coefficient processing, thereby effectively improving the image fusion effect. Considering different fusion strategies for wavelet coefficients and scale coefficients respectively can retain the detail information of the image to the greatest extent, while effectively improving the visual effect of the synthesized image, thereby effectively enhancing the quality of the fused image.

Claims

1. A device for on-site fingerprint search and evidence collection, characterized in that, It includes a moving housing, an imager, a combined light source, a combined light source lifting mechanism, and a controller; The imager and the combined light source are respectively arranged inside the moving housing. The imager is located above the combined light source. The combined light source includes a light source base, a mercury lamp, and a multi-spectrum LED light source. The mercury lamp and the multi-spectrum LED light source are respectively fixed on the light source base; The combined light source lifting mechanism and the controller are respectively fixed to the moving housing. The combined light source lifting mechanism is used to drive the combined light source to rise and fall. The combined light source lifting mechanism includes a lifting stepper motor, a main gear, a pair of slave gears, and a pair of wire reels. The output end of the lifting stepper motor is connected to the main gear. The pair of slave gears are respectively located on the opposite sides of the main gear and mesh with the main gear. The pair of wire reels are coaxially arranged with the pair of slave gears respectively. A suspension wire is wound on each wire reel. The suspension wires on the pair of wire reels are respectively connected to the light source base; The signal input end of the controller is connected to the signal output end of the imager. The output end of the controller is connected to the control input end of the combined light source lifting mechanism. The controller has a display screen to display the images collected by the imager; The controller is used to perform image processing on multiple groups of image signals sent by the imager and display the images after image processing. The image processing includes geometric registration, feature extraction, feature optimization, and feature fusion.

2. The on-site fingerprint search and evidence collection device according to claim 1, characterized in that, The mercury lamp is a ring-shaped mercury lamp, and the ring-shaped mercury lamp is a 254nm short-wave ultraviolet lamp. The multi-spectrum LED light source is a ring-shaped multi-spectrum LED light source. The ring-shaped mercury lamp is located on the periphery of the ring-shaped multi-spectrum LED light source. The ring-shaped multi-spectrum LED light source is composed of a long-wave ultraviolet LED lamp, a violet LED lamp, a blue LED lamp, a green LED lamp, and a near-infrared LED lamp.

3. The on-site fingerprint search and evidence collection device according to claim 2, characterized in that, The long-wave ultraviolet LED lamp, the violet LED lamp, the blue LED lamp, the green LED lamp, and the near-infrared LED lamp are respectively composed of 4 long-wave ultraviolet LED lamp beads, 4 violet LED lamp beads, 4 blue LED lamp beads, 4 green LED lamp beads, and 4 near-infrared LED lamp beads. The long-wave ultraviolet LED lamp beads, the violet LED lamp beads, the blue LED lamp beads, the green LED lamp beads, and the near-infrared LED lamp beads are arranged alternately. Among the 4 LED lamp beads of each color, the connection line between 2 of the LED lamp beads passes through the center of the circle, and the connection line between the other 2 LED lamp beads also passes through the center of the circle. And the connection line between 2 of the LED lamp beads is orthogonal to the connection line between the other 2 LED lamp beads.

4. The on-site fingerprint search and evidence collection device according to claim 1, characterized in that The controller is a computer.

5. The on-site fingerprint search and evidence collection device according to claim 2, characterized in that The imager includes a lens, a filter, and a CCD image sensor arranged in sequence from bottom to top.

6. The on-site fingerprint search and evidence collection device according to claim 5, characterized in that, The imager includes a 550nm long-pass filter, a 580nm long-pass filter, a 600nm long-pass filter, and a filter switching mechanism. The filter switching mechanism is used to selectively move one of the multiple filters to between the lens and the CCD image sensor under the control of the controller; The on-site fingerprint search and evidence collection device further includes a light source driving circuit, which is used to control the turning on and off of the combined light source under the control of the controller.

7. A forensics method using the on-site fingerprint search and forensics device as described in claim 1, characterized in that It includes the following steps: S1. The on-site fingerprint search and evidence collection device receives a power-on instruction, turns on the imager, and enables the display screen to display real-time images. S2. Align the searched on-site fingerprint by moving the moving housing of the on-site fingerprint search and evidence collection device. S3. Make the on-site fingerprint search and evidence collection device work in the reflection mode first, and then in the fluorescence mode. When the on-site fingerprint search and evidence collection device is in the reflection mode, the following steps are sequentially executed: S31. The combined light source lifting mechanism drives the combined light source to move to a preset highest position, and then sequentially turns on the short-wave ultraviolet lamp, long-wave ultraviolet LED lamp, violet LED lamp, blue LED lamp, green LED lamp, and near-infrared LED lamp, and respectively obtains 6 fingerprint photos by imaging. S32. The combined light source lifting mechanism drives the combined light source to move to a preset intermediate position, and then sequentially turns on the short-wave ultraviolet lamp, long-wave ultraviolet LED lamp, violet LED lamp, blue LED lamp, green LED lamp, and near-infrared LED lamp, and respectively obtains 6 fingerprint photos by imaging. S33. The combined light source lifting mechanism drives the combined light source to move to a preset lowest position, and then sequentially turns on the short-wave ultraviolet lamp, long-wave ultraviolet LED lamp, violet LED lamp, blue LED lamp, green LED lamp, and near-infrared LED lamp, and respectively obtains 6 fingerprint photos by imaging. When the on-site fingerprint search and evidence collection device is in the fluorescence mode, the following steps are sequentially executed: S34. The filter mirror switching mechanism moves the 550nm long-pass filter mirror between the lens of the imager and the CCD image sensor. The combined light source lifting mechanism drives the combined light source to move to the preset intermediate position, and then sequentially turns on the short-wave ultraviolet lamp, long-wave ultraviolet LED lamp, violet LED lamp, blue LED lamp, and green LED lamp, and respectively obtains 5 fingerprint photos by imaging. S35. The filter mirror switching mechanism removes the 550nm long-pass filter mirror, moves the 580nm long-pass filter mirror between the lens of the imager and the CCD image sensor, and then sequentially turns on the short-wave ultraviolet lamp, long-wave ultraviolet LED lamp, violet LED lamp, blue LED lamp, and green LED lamp, and respectively obtains 5 fingerprint photos by imaging. S36. The filter mirror switching mechanism removes the 580nm long-pass filter mirror, moves the 600nm long-pass filter mirror between the lens of the imager and the CCD image sensor, and then sequentially turns on the short-wave ultraviolet lamp, long-wave ultraviolet LED lamp, violet LED lamp, blue LED lamp, and green LED lamp, and respectively obtains 5 fingerprint photos by imaging. S4. The controller performs geometric registration, feature extraction, feature optimization, and feature fusion on the 33 fingerprint photos obtained in step S3 to obtain an optimized fingerprint photo, automatically saves the optimized fingerprint photo, and displays it on the display screen.

8. The evidence collection method according to claim 7, wherein In step S3, the on-site fingerprint search and evidence collection device receives a one-key evidence collection instruction and automatically works in the reflection mode first, and then in the fluorescence mode.

9. The evidence collection method according to claim 7 or 8, characterized in that, The feature fusion includes the following steps: Perform non-downsampled morphological Haar wavelet transform on the 33 fingerprint photos obtained in steps S31 to S36 respectively to obtain the low-frequency coefficients at the total scale number J and the high-frequency coefficients at each scale j (0 < j ≤ J); Fuse the low-frequency coefficients according to the fusion rule of the low-frequency coefficients, and fuse the high-frequency coefficients according to the fusion rule of the high-frequency coefficients; Perform consistency detection on the obtained low-frequency fusion coefficients and high-frequency fusion coefficients respectively, and use the detection results to update the low-frequency fusion coefficients and high-frequency fusion coefficients; Perform the inverse transform of the non-downsampled morphological Haar wavelet transform on the final low-frequency fusion coefficients and high-frequency fusion coefficients to obtain the final fused image.

Images