Image processing virtual circuitry for physics based applications on field programmable gate array technology

Abstract

Presented in this paper are the designs of several algorithms which enable the identification and tracking of various regions within a series of images using FPGA technology. Two example probLem domains in the areas of plasma physics and observational astronomy have been expolored. In the plasma physics application, an initial pixel extraction technique has been expanded to include spatial (related to the distance from a cathode) and value measurements (based on intensity values), which are subsequently merged in order to identify different intensity / emissivity regions within a plasma assisted deposition system. The optimum combination of these and other techniques are discussed, together with their reasons for selection. The control signals, decoding method and subsequent processing steps, associated with how a point within individual images is selected, will also be presented. In the astronomical application, a variation of a scene change detection mechanism is shown and how this system was adapted in order to track and chart the motion of Near-Earth Objects (NEOs).

Vector analysis for direction prediction on image strings

“…This work continues the work presented in [1] where a tracking system was developed by combing several images into each other in order to create a visible record or "wake" of where an object had been. The latter is similar to multiple exposure photography where two or more images are superimposed in order to create a single image.…”

“…This system will extract trajectory features, combine and merge them with previous image, in order to identify the vectors and translate them into a bearing [6]. The end result is that it would hopefully decompose a video sequence into temporal segments which would then be analysed, similar to how this work uses a string of images as in [1]. As motion is a critical part of this system [8], if no motion is detected, then a '0' is output from the 'Direction_bearing' signal.…”

“…The design of the system will follow on from the work presented in [1] and [4] and will share some common features from the work already mentioned in Section 1. This system will extract trajectory features, combine and merge them with previous image, in order to identify the vectors and translate them into a bearing [6].…”

“…In Section II the design of the tracking system based upon modifications to the system discussed in [1] is presented. In Section III, the explanation of how this is implemented in DSP Builder is explained, whilst the resultant MATLAB interface and concluding statements are presented in Section IV.…”

Upgrading to a soft multifunctional image processor for implementation on a field programmable gate array with additional biasing and logical capabilities

“…Due to the increased number of operations that provide more control command possibilities, the bus widths of the control signals had to be expanded to compensate for this. The logical arrays shown along the left side of Figure 1 (the 3 signals that go into the AND gate) have been removed in Figure 2 to save resources as in [14]. This helps to reduce the amount of space used up in order to allow the soft morphological operations to be added later.…”