Embedded Multispectral Image Acquisition Node Design

With the development of computer and network technology, remote video monitoring based on embedded system has been applied to agricultural planting,breeding and other fields, which provides real-time, accurate image information for the production management and improves labour productivity greatly. Multispectral image analysis plays an important role in moisture detection, nutritiondiagnosis and other aspects. But the traditional image monitoring system only records the visible range of image information but lost the ultraviolet, near-infrared, mid-infrared bands and other image information which is the important analysis about the composition and growth status of crops.We improved the previous image node from the aspects of high resolution multispectral image acquisition, four bandsswitch automatically, wireless acess and embedded environment support.In the design of the node, we used the imaging dector structure of filter conversion and CMOS with ARM-Linux platform, and we designed some modules which in the layer of kernel hardware driver and user application. The node can acquisite the resolution of 2592×1936(5.0M) pixel color, green(550nm), red(650nm), near infrared(800nm) four kinds of image and wireless transmission. In order to apply nodes conveniently, we designed an agricultural multispectral image acquisition system, through whose client-side user can control the nodes which deployed in the field.The multispectral image acquisition test showed that the collection image’s resolution was 2592×1936(5.0M) pixel, the size of image file was about 1410 KB, the average time of acquisition and compression was 15.9 seconds; Two adjacent spectral bands switch took 2 second, the complete collection(four) spectral image took 70 secods average; The analysis of spectral image quality indicates that the node and MS-4100 cameras in the green, red and near infrared spectral image under the three bands of histogram matching is 0.89, 0.88, 0.85, both multispectral imaging capability gap is smaller. The network transmission test showed that node network instruction loss rate was 1%, the 3G network transfer rate was about 92 Kbps, send an image need time between 15 seconds to 25 seconds. Through the preliminary application of plant chlorophyll detection, it displayed that the node in the agricultural scientific research has a broad application prospect.