Research on intelligent lighting control system based on wavelet transform and image fusion

1 Introduction

Since the beginning of the 21st century, my country's buildings have entered an era of highly intelligent development. New intelligent buildings and modern residential communities can no longer meet their higher standards with traditional lighting control methods. Traditional lighting methods are simple, effective, and intuitive, but they rely too much on the personal ability of the controller, the control is relatively decentralized and cannot be effectively managed, and its timeliness and automation are too low. Although the subsequent automatic lighting control mode solves the problems of relatively decentralized control and ineffective management of traditional methods and realizes the automation of lighting control, it cannot realize the dimming control function.

At present, foreign products such as Niko's intelligent lighting control system can preset various scenes for lighting control and are widely used in office buildings, hotels, stadiums and other places, but they have disadvantages such as high price, relatively complex operation, and high requirements for management personnel. There are still few related lighting intelligent control systems used in residential areas and general public places in China.

In order to solve the shortcomings of the traditional infrared + light sensor lighting control system, such as requiring more sensors, high requirements for layout positions, large engineering construction and wiring, this paper proposes an illumination control method that uses dynamic and static monitoring (infrared, voice control) + digital image information fusion. The collected image information is fused, the fused image is divided into grayscale, and the grayscale average value of each area is compared with the preset grayscale value to adjust the scene illumination.

2. Fusion technology of motion and static monitoring sensor data and CCD digital image information

The control system block diagram of the dynamic and static monitoring and CCD digital image information fusion technology in the building intelligent lighting system is shown in Figure 1.

The basic principle is: observe whether there is anyone walking through motion detection technology. If no one is walking, turn off the lighting; if there is someone, analyze the multiple digital images collected, compare the image grayscale average with various preset standard values, and calculate the illumination model of the ambient illumination field. If it is within the error range allowed by the preset mode of the lighting system, there is no need to adjust the illumination, otherwise, the illumination needs to be adjusted.

Image fusion refers to the process of obtaining a composite image by using a certain fusion technology after denoising, temporal registration, spatial registration and resampling of images of the same scene obtained by different sensors or the same scene obtained by the same sensor at different times. By fusing multiple sensor images, the limitations and differences in geometry, spectrum and spatial resolution of a single sensor image can be overcome, and the image quality can be improved, which is conducive to the location, identification and interpretation of physical phenomena and events. The specific process is shown in Figure 2.

3 Image fusion technology based on wavelet transform

3.1 Preprocessing

In the intelligent lighting system, when the motion monitoring finds someone walking, the CCD camera will collect image information of the corresponding area. However, during the image acquisition process, due to the influence of various factors (such as the position speed of the sensor, light intensity, random noise, etc.), the actual image often contains the characteristics of the above factors. Therefore, before realizing image fusion, it is necessary to pre-process the different images obtained by the sensor, including image correction, enhancement, smoothing, filtering, registration, etc.

As shown in Figure 3, the indoor image captured by the CCD camera is displayed after image correction, filtering and registration preprocessing.

It can be seen from Figure 3 that due to factors such as light intensity, noise and interference, the flower pot on the right in Figure 3 (a) is a little blurry, and the door in Figure 3 (b) is a little blurry. The information provided by such images is not conducive to the recognition of the intelligent lighting system. Therefore, the following wavelet fusion technology can be used to fuse the source image information.

3.2 Wavelet transform fusion

Inspired by Burt and Adelson's pyramidal image decomposition and reconstruction algorithm, Mallat proposed the Mallat fast algorithm for wavelet transform. According to the two-dimensional Mallat algorithm, each preprocessed CCD image is decomposed into two dimensions.

The image size collected by the CCD camera in this paper is 351 × 260. The number of decomposition layers is set to 3, and the decomposition is performed at scale k-1 according to the following Mallat decomposition formula:

In the formula,

They represent the low-frequency component, horizontal high-frequency component, vertical high-frequency component and diagonal high-frequency component of the preprocessed CCD image at a resolution of 351 × 260. The low-frequency component reflects the approximate and average characteristics of the CCD image and concentrates most of the energy information of the image. As shown in Figure 4, it is a schematic diagram of wavelet decomposition of the CCD image.

In the wavelet transform domain of the two CCD images, the horizontal, vertical and diagonal components are fused respectively. The high-frequency coefficients of the two CCD images are compared at each scale j (j = 1, 2, 3), and the coefficients with larger absolute values ​​at the corresponding positions are retained as important wavelet coefficients, that is,

Respectively represent the wavelet coefficients of the two CCD images at each scale component.

The approximation coefficients C1J and C2J of the two CCD images after wavelet transform are processed. Due to various factors when the intelligent lighting system collects images, the images collected by the CCD camera are blurred in some parts. The blurred image means that its detail information (or high-frequency information) is lost more. In contrast, its overall information (or low-frequency information) is better preserved. Therefore, the difference between the approximation coefficients of the two CCD images after wavelet decomposition is much smaller than the difference between the wavelet coefficients. Therefore, the approximation coefficient after fusion can be determined.

Using all the wavelet coefficients obtained above and the approximation coefficients of multiple CCDs collected by the intelligent lighting system, a two-dimensional wavelet inverse transform is performed, and a reconstruction-type image fusion image is obtained. The fusion process is shown in Figure 5.