DE10053180A1 - Building monitoring system has camera output connected via wavelet encoder to device for boosting sharpness - Google Patents

Abstract

The system includes a camera (10) for outputting an image as an output signal. A wavelet encoder (12) encodes an output signal from the camera. A wavelet decoder (14) may also be included. A device (16) boosts the image sharpness after wavelet decoding. A memory (18) stores the output signal from the wavelet encoder and/or the wavelet decoder and/or an output signal from the device (16) for boosting image sharpness. An Independent claim is included for a method of operating a building monitoring system.

Description

The invention relates on the one hand to a house surveillance system with a camera, which is designed to take a picture and an output corresponding to the picture signal.

Such home surveillance systems are known. The camera is often on a monitor connected, the monitor the image taken by the camera reproduces. The image data contained in the output signal from the camera is very high extensive, which is why each of the direct reproduction of the image using a monitor further processing or processing and use of the recorded by the camera my pictures is difficult.

The invention therefore has the task of a home surveillance system type mentioned in such a way that further processing, further processing tion and / or other use of the images is simplified.

According to the invention, the task is performed by a wavelet encoder Encoding an output signal from the camera solved.

The invention is based on the knowledge that the wavelet coding a reduction in the amount of data of 100: 1 and more can be achieved, so that a Further processing, further processing or other use of the image data is possible, without having to deal with too much data.

In the event that the image data encoded with the encoder is restored According to the invention, a wavelet decoder is preferably provided.

With wavelet coding with subsequent decoding, the amount of data - at least temporarily - greatly reduced, as has already been stated above. is the compression rate, which is preferably adjustable, too high, can ver image information lornen, which lead to blurring. A device is therefore preferred according to the invention device for increasing the sharpness of an image after wavelet decoding hen. This allows an image to be created that is no longer all from the camera captured information includes, but for the purposes of home surveillance without further is sufficient.

The further processing, editing or other use of the image is in principle ell any. According to the invention, however, a first memory for storing one is preferred Output signal of the wavelet encoder and / or the wavelet decoder and / or one Output signal of the device for increasing the image sharpness provided, in the inter can be saved to prepare for further steps.

In addition, a second memory for storing comparison image data and a comparison device for comparing the comparison image data with

• a) an output signal of the wavelet encoder,
• b) an output signal of the wavelet decoder and / or
• c) an output signal of the device for increasing the image sharpness is provided his.

This makes it possible to preview the pictures taken by the camera to compare stored images, which, for example, monitors the users Well-known people from the house at the front door of the house with the camera can be recognized.

In this context, an off can then be preferred according to the invention value circuit for initiating a predetermined process depending on the result of the comparison carried out by the comparison device. That's the way it is For example, it is conceivable that the residents of the house will be recognized at the front door whereupon the door opens automatically. Even with people who may not belong to the household but are known to members of the household, another process than people who do not belong to this circle.

According to the invention, a connection for remote data transmission can preferably be used for this purpose be provided. For example, if an acquaintance of a household member the doorbell rings, appropriate information such as "Mr. Müller is there" transmitted to a household member's cellular phone via WAP. Because of the Data compression by wavelet coding described above can also be used even the image data on the radio telephone or on a corresponding monitor outside of the house. The Wavelet- Decoding and / or the focus increase can be performed.

According to a further preferred embodiment of the invention, a Motion detector provided. This motion detector can be installed in such a way that he registered and given movements within the recording area of the camera if the camera turns on. As a result, the system only has to be operated if  something actually changes in the field of view of the camera. These applications are particularly particularly advantageous with regard to possible break-ins, of course the Ka mera then not necessarily in the entrance area, but in addition or alternatively to others locations where the house is at risk of break-ins.

In addition to the further processing and processing of the image data described above it goes without saying that the image can also be displayed directly. Therefore, a is preferred Monitor for displaying an output signal of the wavelet decoder and / or the Device provided for increasing the image sharpness corresponding.

In addition to the home surveillance systems mentioned at the beginning, there are also procedures known to operate such home surveillance systems.

These processes are also in part because of the image processing and processing as well as the amount of data to be managed with other uses.

The invention is therefore based on the additional object of a method for Operate a home surveillance system with a camera that is designed to Take picture and output an output signal corresponding to the picture, such to train that further processing, further processing and the other groove tion of the image data is simplified.

According to the invention, the object is achieved in that an output si gnal of the camera is compressed by wavelet coding.

In other words, wavelet coding is used to deal with them to drastically reduce the image data, for example by a factor of 100: 1 or more.

It can be provided according to the invention for the recovery of the images that an encoded output signal is decoded (again).  

Since u. U. in the coding and subsequent decoding of the image data Informati Ons losses can occur, which lead to a reduction in sharpness, is invented According to the invention, it is preferably provided that the image sharpness of an image according to Wavelet-De coding is raised.

With regard to the diverse possibilities of further processing and processing as well The other use of the image data is preferred according to the invention that a coded Output signal and / or a decoded output signal and / or after an image Raising the image sharpness corresponding signal is / are stored.

Such storage can be done, for example, for documentation purposes a so-called endless loop, so that, for example, always the first last hour by saving images in the facial area of the camera are documented.

A preferred possibility according to the invention for further processing and / or processing of the image data is that

• a) an encoded output signal before decoding,
• b) a coded output signal after decoding and / or
• c) a signal corresponding to an image after increasing the sharpness of the image

is / are compared with a previously stored signal.

The comparison can be carried out at regular intervals. This Ausge staltung is particularly in surveillance cameras to protect against break-ins Advantage because it detects every change in the field of view of the camera can.

Especially when the comparison serves the identity of a person To determine how this has already been explained above, the comparison can also be appealing  be carried out when the camera is switched on. Turning on the camera in such a case, it is then preferably coupled to an output signal of a accordingly arranged motion detector and / or by pressing the bell button on Entrance.

As already explained above, the invention can provide that dependent a predetermined process is initiated from the comparison result. If the A comparison, for example, could show that a burglar is approaching the house an alarm is triggered.

The invention is explained in more detail below with the aid of a preferred exemplary embodiment. The only figure shows
a block diagram of an embodiment of the home surveillance system according to the invention.

According to the block diagram, a camera 10 is provided, which can be hung, for example, in the area of the house entrance. However, it can also be located in other places in the house, for example in particularly sensitive areas. Of course, several cameras can also be provided.

The output signal of the camera goes to a wavelet encoder 12 . This encoder performs a so-called wavelet transformation (WT). Similar to the Short Time Fourier Transformation (STFT), this is an orthogonal transformation that is continuous in the signal and spectral range:

The WT is defined by the convolution of the signal x (t) with a model bandpass of the center frequency f ₀ and the impulse response γ (t). The time variable t in the above equation can be replaced by the location variables r and s, and the above equation can also be extended to the more dimensional case. As with the STFT, the output signal of the WT is dependent on both the frequency and the time of analysis. The limited impulse response γ (t) of the model bandpass is, however, increasingly compressed on the time axis with increasing analysis frequency f. This results in a bandpass with f as the center frequency, but which also has an impulse response length that is inversely proportional to f. It is therefore obtained at low frequencies a very high resolution in the frequency range, but a low resolution in the signal range, while at high frequencies there is a very low resolution in the frequency range, but a high resolution in the signal range. The bandpass model function of the WT, γ (t) (referred to as the wavelet) is to be orthogonal to its compressed version, just as in the short-term Fourier transformation, the analyzing e-functions are to be converted components of an orthogonal transformation by modulation. Bandpass analysis filters that result from modulation of a model lowpass filter therefore also lead to orthogonal subband signals.

A discrete wavelet transformation (DWT) can also be carried out. The number of frequency bands is finite:

• - In contrast to the classic transformations, the DWT does not use any Basic functions of constant length; the length of the impulse responses behaves a lot more reciprocal to the bandwidth of the analyzed frequency range.
• - At low frequencies, the values in the signal range are strongly undersampled, weh rend the bandwidth is small; the product of bandwidth and subsampling factor is constant in all frequency bands.

In practical terms, the DWT is nothing more than a subband in octave bands. The complexity of DWT can increase with frequency ever shorter filter lengths never become larger than twice the computation length of the lowest frequency baseband necessary complexity. Subband process with non-uniform subband decomposition therefore offer a more efficient realization of the Fre Frequency analysis as linear transformations, which unnecessarily use the high-frequency Display parts with high frequency resolution.

The so-called hair transformation is also called hair wavelet transformation denotes and actually represents the simplest possible realization of the DWT, whereby  Realizations with filters with a longer impulse response - through a different choice of the γ (t) function - however, allow a sharper spectral separation of the frequency bands.

Regarding the design of the filters used, the following should be noted: filters, which are orthogonal to their modulated versions do not necessarily have to be ska versions of themselves are orthogonal. In fact it takes place in cascaded part band systems a multiple series connection of one and the same filter instead, wherein due to the undersampling and oversampling, the filter impulse response is scaled, i. H. is compressed or stretched by a factor of 2. Filters that both against their modu gated, as well as orthogonal to their scaled versions, are called biortho designated gonally.

The specification of an octave band division in the WT represents a certain one Limitation that may not be optimal for frequency decomposition of any type of signal got to. Wavelet packet analysis is therefore a variant of the DWT in which the analysis bandwidth can be selected in almost any frequency range, d. H. Depending on the requirements, high-pass analysis branches can also go further individually be disassembled. This means that both the STFT (in the discrete case: block and block overlap pende transformations) as well as the WT / DWT special cases of the wavelet packet analysis represents.

After the wavelet coding in the encoder 12 , in which the compression can be predetermined by appropriate setting of the parameters (for example the coefficients), a wavelet decoding takes place in a wavelet decoder 14 . Depending on the compression selected, the information content of the encoded and de-encoded image may have been lost, which in particular can affect the sharpness of the image. For this reason, an image sharpness increase takes place in an image sharpener 16 . In the simplest case, this can be done by the sharpness enhancer 16 influencing the wavelet coefficients, in particular enlarging them. In the simplest case, this is done by multiplying the wavelet coefficients by a factor. However, the factor must be adapted to the respective image content. This method is therefore particularly suitable when an image does not necessarily have to be processed in real time, but when there is sufficient time to determine the optimal factors for the wavelet coefficients.

Another way to increase the focus is to differentiate the smoothed image:
First of all, a low-pass filter T or a high-pass filter H is required, the amplification for DC components

is. With the help of this filter, the original signal x (n) is first filtered line by line (without subsampling), ie for each pixel n with n = 0. . (N _H N _V - 1), N _H width of the image in pixels, N _V height of the image in pixels, a filtered pixel y (n) is generated.

In the case of a high-pass filter, these coefficients y (n) are regarded as the "wavelet coefficients" d (n) of the subband HG. However, if the selected filter is a low-pass filter, the "wavelet coefficients" d (n) are formed as a difference from the original signal x (n).

d (n): = x (n) - y (n) (2)

The coefficients d (n) are now, using a given formula for the HG band, stored in the corresponding storage position. That memory area is four times the size of the original image.

The previously calculated coefficients d (n) must now be repeated, but now column by column, filtering for the coefficients of the GG band. To do this the coefficients d (n) of the previous step as the coefficients x (n) in equation (1) is viewed and again according to equation (1) and possibly according to equation (2) method. The resulting coefficients d (n) are called the "wavelet coefficients" of the GG band using the given form (for the GG band to the corresponding appropriate memory position in the four times larger memory area.

The coefficients of the subband GH are now missing, which can be High pass filtering occurs when you go through the original image column by column. The ent standing coefficients are ge at the storage location for the subband GH wrote.

Finally, the original image is written into the memory as subband HH.

The sharpened signal is obtained by the inverse wavelet transformation, whereby make sure that a symmetrical synthesis wavelet is used. The recon structured image is then an image with double pixel density. To get a sharper picture of size To get the original picture, you have to "downscale" the reconstructed picture ren ". Since only one" transformation stage "was carried out, the memories correspond positions of the HH coefficients in the reconstructed image every second pixel, with what the image can be transformed down to half the size.

Of course, this process can also be used to sharpen color images. If the color image in its three components RGB is the required storage space three times the size of grayscale images. Furthermore, the coefficients for the three subbands HG, GH and GG can be calculated for all three color separations. The synthesis must also be carried out for the three large memory areas, which is a ver wasted computing time.  

However, if the color image is transformed from RGB to YCrCb, a can simplified procedure can be applied. Here, the above algorithm is only applied to the Luminance component (brightness component, Y band) applied. The additional memory requirement and the computing time is then just as long as for grayscale images. However, this leads to small quality losses in the sharper picture. Following the synthesis and that "Downscaling" will then be the back transformation from YCrCb to RGB with the old color difference values performed.

An image data memory 18 is connected to the output connection of the sharpness enhancer 16 . The image data memory 18 has to provide considerably less storage space than would have been the case if the image data from the camera 10 had been stored directly, ie without wavelet coding, that is to say uncompressed.

As already mentioned above, the image data stored in the image data memory 18 can be compared, for example by means of a comparator 20, with image data previously stored in a reference image memory 22 (biometrics). For example, data about pictures of household members can be stored as reference image data. If the comparison carried out by the comparator 20 then shows, for example, that the person picked up by the camera 10 at the house entrance belongs to the household, the comparator 20 gives an impulse to a door opener 24 so that the person belonging to the household can enter without to use a key or the like.

Of course, it is not absolutely necessary that only images after the focus has been raised are stored in the image focus memory 16 in the image data memory 18 . Rather, it is of course also conceivable for image data to be stored in the encoder 12 in the first memory 18 immediately after wavelet coding and to be compared with correspondingly coded reference images. In addition, this comparison can also take place after decoding in the decoder 14 , but bypassing the sharpness of the image in the sharpener 16 . The comparison can also take place in real time, ie bypassing the memory 18 .

The image decoded in the decoder 14 can also be shown with or without a prior sharpness increase in the sharpness enhancer 16 by a monitor 26 , which is advantageously located inside the house.

The encoder 12 , the decoder 14 , the sharpness enhancer 16 , the image data memory 18 , the comparator 20 and the reference image memory 22 are all realized in the illustrated embodiment in the context of a single digital signal processor (DSP) 28 .

Of course, the images recorded by the camera 10 can also be displayed directly in the monitor 26 .

The camera 10 is controlled via a motion detector 30 which is arranged in the region of the field of view of the camera 10 .

In addition to the application described above, there are various other possible applications for the present invention, as is realized in the exemplary embodiment according to the attached block diagram:
For example, reference images of acquaintances of the operator of the system can be stored in the reference image memory 22 . If the camera 10 takes an image after activation by the motion detector 30 and determines that the image shows a friend whose image is stored in the reference image memory 22 , then the comparator 20 initiates a call to a preset telephone number and on a telephone connection 32 transmits, for example, the "WAP" message: "Mr. Meier is here".

The comparison in the comparator 20 can be carried out on the basis of the output signals of the encoder 12 , the decoder 14 and / or the sharpness enhancer 16 , with or without intermediate storage in the image data memory 18 , depending on the form in which the reference images in the reference image memory 22 are filed.

Instead of the WAP message, image data can of course also be transmitted, either in the form of the output signal of the encoder 12 , which means that the least amount of data has to be transmitted, in the form of the output signal of the decoder 14 , ie before the focus is raised and in the form of the output signal of the image sharpener fenanhebers 16 . Should the output signal of the encoder 12 be transmitted, a corresponding decoder must of course be provided on the receiver side. If the image sharpness of the decoded signal is too low, an image sharpness enhancer would also have to be provided on the receiver side.

The output signals of the encoder 12 , the decoder 14 , the image sharpener 16 and the comparator 20 can also go to a homepage server 34 , from which they can then be retrieved by a user, if appropriate after appropriate authorization. The content of the image data memory 18 could also be called up via the server 34 . Again, appropriate means must be provided on the part of the retriever for processing the retrieved images, provided that the output signal of the encoder 12 or the decoder 14 is retrieved.

Instead of using the motion detector 30 , which according to the block diagram is used to switch on the camera 10 , the monitor 26 , the comparator 20 and / or the homepage server 34 , said switching on can also be carried out by pressing a bell button (not shown).

The image data (encoded, decoded, with and / or Focus increase) are saved periodically, so that, for example, always the Image data from the last hour are recorded. This function is particularly useful when the camera is used to monitor for possible break-ins. In particular when evaluating such images, the focus increase can also be used by manipulating the wavelet coefficients because in this case there is enough time is available to determine the optimal coefficients.

Those disclosed in the above description, the claims and the drawing Features of the invention can be used both individually and in any combination  the realization of the invention in its various embodiments essential his.

Claims (17)

1. Home surveillance system with
a camera ( 10 ) which is designed to take an image and to output an output signal corresponding to the image,
marked by
a wavelet encoder ( 12 ) for encoding an output signal of the camera. 2. Installation according to claim 1, characterized by a wavelet decoder ( 14 ) 3. Installation according to claim 2, characterized by a device ( 16 ) for increasing the sharpness of an image after the wavelet decoding. 4. System according to one of the preceding claims, characterized by a first memory ( 18 ) for storing an output signal of the wavelet encoder ( 12 ) and / or the wavelet decoder ( 14 ) and / or an output signal of the device ( 16 ) for lifting the sharpness. 5. Plant according to one of the preceding claims, characterized by
a second memory ( 22 ) for storing comparison image data and
a comparison device ( 20 ) for comparing the comparison image data with

• a) an output signal of the wavelet encoder ( 12 ),
• b) an output signal of the wavelet decoder ( 14 ) and / or
• c) an output signal of the device ( 16 ) for increasing the sharpness of the image.

6. System according to claim 5, characterized by an evaluation device for initiating a predetermined process as a function of the result of the comparison carried out by the comparison device ( 20 ). 7. Installation according to one of the preceding claims, characterized by a connection ( 32 ) for remote data transmission. 8. Installation according to one of the preceding claims, characterized by a motion detector ( 30 ). 9. System according to one of claims 2 to 8, characterized by a monitor ( 26 ) for displaying an output signal of the wavelet decoder ( 14 ) and / or a device ( 16 ) for increasing the image sharpness corresponding image. 10. A method for operating a home surveillance system with a camera ( 10 ) which is designed to take an image and to output an output signal corresponding to the image, characterized in that an output signal of the camera ( 10 ) is compressed by wavelet coding. 11. The method according to claim 10, characterized in that a coded Output signal is decoded. 12. The method according to claim 11, characterized in that the sharpness of an image Image after wavelet decoding is raised. 13. The method according to any one of claims 10 to 12, characterized in that a encoded output signal, a decoded output signal and / or a picture after Raising the image sharpness corresponding signal is / are stored. 14. The method according to any one of claims 10 to 13, characterized in that

• a) an encoded output signal before decoding,
• b) a coded output signal after decoding and / or
• c) a signal corresponding to an image after increasing the sharpness of the image

is / are compared with a previously stored signal. 15. The method according to claim 14, characterized in that the comparison in is carried out at regular intervals. 16. The method according to claim 14 or 15, characterized in that the comparison is carried out in response to switching on the camera ( 10 ). 17. The method according to any one of claims 14 to 16, characterized in that depending on the comparison result, a predetermined process is initiated.